▶ 0:04:20The committee will please come to order. A quorum is present. The committee meets today pursuant to notice. Without objection, the chair may recess the committee at any point. Artificial intelligence is uh scaling rapidly. And as a result, it is increasingly how we learn, how we work, and how we prepare for the future.
▶ 0:04:48Today's hearing examines AI and its growing role in education, workforce development, and the the future of work Artificial intelligence is often described as a sudden disruption, but many of its underlying technologies have existed for decades. What has changed is the scale and the accessibility of these tools.
▶ 0:05:12Rapid advancements have made AI systems easier to use and more widely available, reaching classrooms, workforce training and workplaces across the country. AI's potential benefits are significant. AI tools are already helping teachers reduce administrative burdens, better personalize learning, and free up more time to engage directly with students.
▶ 0:05:38For workers, AI can help build new skills, increase efficiency, and accelerate on-the-job As a result, employers can expect productivity gains that contribute to economic growth. An estimated 40% of GDP could be significantly affected by generative AI, which could mean a permanent increase in economic activity.
▶ 0:06:07That said, AI growth is not without challenges. Models can make mistakes. In education, that raises concerns about academic integrity and over-reliance.
▶ 0:06:20In the workplace, early evidence suggests AI may change how tasks are performed, particularly for entry-level roles, which underscores the need for strong pathways into the labor market and strong skills It's also important to recognize that many labor laws and regulations have not kept pace with how quickly work is and they too often make it harder for businesses to offer benefits and
▶ 0:06:51With the rapid pace of technological change, workers are seeking greater freedom and flexibility in how they This highlights the need for clearer standards for independent work and more flexible benefit models to better serve a modern, mobile workforce.
▶ 0:07:09Today, more than 70 million workers operate as self-employed independent contractors, yet the federal government too often does not provide the statu- statutory clarity that these workers need and deserve. In our current age of rapid technological growth, including AI-related development, our policies must adapt to the growing role of independent work.
▶ 0:07:37As this committee considers AI, we must recognize that many existing education and workforce policies were designed for a different era. The goal should not be to rush into sweeping new rules and regulations, but to ensure schools, employers, and training providers can keep pace with innovation while maintaining trust and prioritizing safety and privacy.
▶ 0:08:04US leadership in AI is pivotal in to the nation's economic success and security. The Trump administration understands that this leadership requires an AI-ready workforce supported by strong classroom instruction and on-the-job learning that make the most of cutting-edge Today, we have the opportunity to better understand how AI is being is being and could be used in education and workforce systems,
▶ 0:08:36where it's delivering value and where challenges remain. I look forward to hearing from our witnesses, and with that, I yield to the ranking member. Thank you, Mr. Thank you, Mr. [clears throat] Chairman, and thank you to our witnesses for being with us
▶ 0:09:00From chatbots to programming to grocery lists, artificial intelligence is rapidly integrated into our daily lives of millions of Americans, transforming how we interact with technology and the world around us. Due [snorts] to the rapid rise of AI technology, scientists and researchers have had relatively little time to study its impact on American lives.
▶ 0:09:22So, we must act carefully and methodically when implementing new technologies, particularly in settings where they may have long-lasting uh impacts, such as our schools, workplaces, and healthcare Uh this is a time of incredible change and possibility, so it is up to us to shepherd this new technology with caution and responsibility that the situation demands.
▶ 0:09:45As we discussed at an early childhood elementary and secondary education subcommittee hearing earlier this year, AI is already present in our schools. Many educators have found that generative AI systems allow them to offer more personalized instruction to students, streamline assignments, and provide real-time feedback. This has led to a rise of AI literacy in which students and teachers have learned to use AI responsibly for the benefit of their education and futures.
▶ 0:10:15But it's not without risks. There's a piece this morning on NPR uh titled "Report the risks of AI in schools outweigh the benefits." Um while so while the prospect of this new technology is exciting, we don't have enough data right now to understand the full impact on Americans' lives. AI tools are created by humans, and humans are flawed.
▶ 0:10:40This can lead to AI tools exhibiting biases and unfairness due to bias bias training data, other kinds of discrimination and lack of diversity in AI development. This can also result in unfair decision-making about what students need to learn and how. Discrimination by way of AI has also had ramifications in our workforce and society.
▶ 0:11:05The Society for Human Resource Management Survey found that nearly 80% of its members use AI for recruiting and Multiple studies have found that some of these recruitment tools have demonstrated biases against women and black applicants. It's important to recognize that AI is not an infallible technology, and it is not neutral.
▶ 0:11:25So, we must take proactive steps to ensure that AI technology does not perpetuate existing disparities and erode civil Concerns about AI in the workplace doesn't stop with the hiring process.
▶ 0:11:38As AI proliferates, we've been begun to see employers use sophisticated automatic technology to monitor and evaluate This includes tracking how workers' physical activities This includes tracking workers' physical activities on the job, such as how often they go to the bathroom, using facial recognition to assess their attention to comp- computer screens, and others.
▶ 0:12:04Still worse, employers have used data to estimate potential union organizing I don't think anyone on this committee would want their personal activities tracked by an employer. So, she would So, we shouldn't expect less from the American for for for the American
▶ 0:12:23Fortunately Unfortunately, the Trump administration's actions to dismantle the Department of Education have hamstrung much of our existing capacity to study and regulate AI's effect on The decimation of the Office of Education Technology and the Institute of Educational Science Sciences has diminished our ability to comprehensively research student safety related to artificial intelligence.
▶ 0:12:49One of our strongest institutions to fight discrimination, the Office of Civil Rights, has had almost half its staff fired and seven of its 12 regional office closed. Now, while the Trump administration has started rehiring some of the staff, the chaos has clearly disrupted the office's operations to the detriment of parents and students. Today, 33 states have released official guidance or policies on the use of AI in K-12 schools.
▶ 0:13:21Virginia, for example, has issued guidance on integrating AI in education and, among other things, provides recommendations for teacher professional development and AI literacy programs. But without clear, consistent federal leadership from the Department of Education, states have largely been left to navigate this complex landscape on their own, creating inconsistencies and exacerbating achievement gaps.
▶ 0:13:47AI also poses significant risks risks to data privacy and child safety. For example, a nonprofit group, AI Forensics, found in a recent analysis of 20,000 images created by Grok, that 2% depicted a person who appeared to be 18 or younger in bikinis or transparent A study by Brown University researchers found that AI chatbots boxes routinely violate mental health ethics standards, including failing
▶ 0:14:18to refer users to appropriate resources and responding indifferently to crisis situations, including suicidal ideation. AI had recently launched a Open AI recently launched a controversial Chat GD GPT health app, which can review users' health records and generate personalized advice, raising concerns about privacy and accuracy of chatbots boxes.
▶ 0:14:47the Meta is currently fighting allegations in court that it downplayed the risks of its platforms to children. The Center for Missing and Exploited also reported seeing a dramatic increase in generative AI technologies that are capable of creating child sexual abuse material and a rise in other disturbing trends related to AI chatbot bots.
▶ 0:15:10The organization received approximately 7,000 reports uh through its CyberTipline in 2024 and nearly half a million in just the first half of 2025. Despite these very real risks, it's become alarmingly clear that the Trump administration is eager to grant large tech technology companies unprecedented access to Americans' data.
▶ 0:15:36In December, the Trump administration issued an executive order attempting to block states from regulating AI on its Um the um Congress seriously considered doing that by legislation, which I think is necessary to override state state laws.
▶ 0:15:55Congress should not, however, stand idly by while the Trump administration may be ingratiating itself to big tech CEOs in preventing states from protecting Americans against dangerous dangers posed by AI. This is why Congress must take a active role in shaping how how AI is integrated into our lives and develop thoughtful regulations that balance protecting students, workers, and and fostering and fostering economic growth.
▶ 0:16:25Today, we stand at another crossroads of technology technological advancement. It's up to Congress to ensure that AI takes us forward, but not backwards. So, I look forward to the conversations today, and yield back the balance of my time. I thank the gentleman. Uh pursuant [snorts] to committee rule 8C, all members who wish to insert written statements into the record may do so by submitting them to the committee clerk electronically in Microsoft Word format by 5:00 p.m.
▶ 0:16:5514 days after this hearing, and without objection, the hearing uh record will remain open for 14 days to allow such statements and other extraneous material noted during the hearing to be submitted for the official hearing record. I'll now turn to our four witnesses. Uh our first witness, uh Ms.
▶ 0:17:18Chaya Nayak, the head of certifications and jobs platform at OpenAI, from San Francisco, California. Welcome. Our second witness is Mr. Adeel Khan, the founder and CEO of Magic School AI, from Boulder, Colorado. Welcome. Our third witness is Ms. Alexandra Reeve Givens, the president and CEO of Magic School AI. Excuse me.
▶ 0:17:52My view shifted there, I guess. Uh Gave you another title. The The president CEO for the Center for Democracy and Technology from Washington, D.C. Welcome. Our last witness is Mr. Kevin Frazier, an AI Innovation and Law Fellow from the University of Texas Law School, from Austin, Texas. Welcome. We thank the witnesses for being here today.
▶ 0:18:15Pursuant to committee rules, I would ask that you each limit your oral uh to a 3-minute summary for our for um for your written statement. The clock will count down from 3 minutes, as committee members have many questions for you, and we would like to spend as much time as possible on questions. Pursuant to committee rule 8D and committee practice, however, we'll not cut you off your testimony until you reach the 5-minute mark.
▶ 0:18:46I would also like to remind the witnesses to be aware that their responsibility to provide accurate to the committee is their And so, I'll first recognize Ms. Nayak for your testimony. Thank you, Chairman Walberg, Ranking Member Scott, and members of the My name is Chaya Nayak, and I lead OpenAI's work on certifications and our jobs platform.
▶ 0:19:14These efforts are focused on helping people build AI skills and connect those skills to real economic opportunity. At OpenAI, we believe that AI will make workers more productive and unlock more opportunity than any technology in It will help people do their jobs better, help businesses operate more and give anyone the power to turn ideas into income and outcomes, and create jobs that don't even exist today.
▶ 0:19:42But we know that work will look quite Some jobs will actually be lost. We want to do our part to help workers actually prepare and get ready for this AI-driven economy. Doing this will require us to acknowledge a simple reality.
▶ 0:19:58These tools can do far more than workers, small businesses, and communities are currently using them for And if we don't close that gap, a select few will capture most of the benefits of AI, while others will be left behind. That gap is what we call the capability overhang. The distance between what AI can do and what people are actually using it for today. Our goal is to close that so AI can work for all Americans.
▶ 0:20:28Employers are being clear about the value they place on AI skills. More than 70% of business leaders say that they would hire a less experienced candidate that has AI experience versus a more seasoned one without it.
▶ 0:20:43And other research shows that workers with AI skills can actually earn up to 50% more than their non-AI skilled At OpenAI, our starting point is ChatGPT For many people, ChatGPT is not just a tool, it's actually a place to learn. Workers use it to practice real tasks, get feedback, and build skills.
▶ 0:21:10And um it's really dynamic and super That learning matters, but only if it helps workers and employers be able to actually acknowledge the skills that they're getting from the use of AI. And this is where our certifications program comes in. It begins with AI Foundations. Um we're building a course that actually helps people learn AI skills and um apply those skills across industries and different outcomes.
▶ 0:21:37With the additional coursework, they're going to progress to a full OpenAI certification, and that broader credential will show employers that a person can actually apply AI effectively in real-world work settings. Our goal is to certify 10 million Americans by 2030 with AI skills. From there, we're working to close the loop with our OpenAI jobs platform. This platform will launch very soon.
▶ 0:22:02The jobs platform is designed to connect employers with job seekers across every industry, and our goal is to use AI to help people get prepared, get matched, and to actually get hired. We're launching our jobs and certifications effort with a broad coalition of partners, employers like Walmart, State Systems, universities, because each of those organizations brings unique insight into what workers need and what employers actually value.
▶ 0:22:30We think there's still time to prepare people for this new AI era, but there's not time to wait and um kind of not build programs and systems that we need to make people successful. That's why at OpenAI, we're acting right now, and we're helping them um by building the tools and technology to support them and make it easier for employers to actually hire them. Thank you so much for the opportunity to testify today, and I look forward to all of your questions. Thank you. I now recognize Mr.
▶ 0:22:57Khan for your uh Uh Chairman Walberg, Ranking Member Scott, and members of the committee. Thank you for the opportunity to testify. My name's Adil Khan. I'm the founder and CEO of Magic School, a generative AI platform built for Before this, I was I spent my career in public schools as a teacher, assistant principal, principal, and district leader. I've held nearly every job in a school building.
▶ 0:23:23There are over 7 million educators using Magic School's AI worldwide, representing nearly every school and district in the US. Teachers are some of the widest adopters of AI in any profession. Many educators are using it in practical ways, like saving time on routine work, lesson planning, personalizing instruction for diverse learners, and communicating with families. That reality is not the problem.
▶ 0:23:47The problem is that we do not yet have shared standards and safe, purpose-built tools as a default for classrooms. Without clear guardrails, responsibility fragments, and districts struggle to protect students and learn what works.
▶ 0:24:01The most promising early uses are AI delivers the most value when it strengthens, not shortcuts, teacher Used well, it helps a teacher prepare faster, differentiate for students, and provide timely feedback, while decisions stay firmly in their hands. K-12 schools are different. Students are minors. Schools hold sensitive data.
▶ 0:24:26Instruction shapes shapes learning Responsible use should be anchored in a few simple principles. Tools purpose-built for education. Humans stay in the loop. Student protections by design.
▶ 0:24:41And transparency so districts and families understand data use and Education policy is primarily state and The most constructive federal role is to support capacity and protections for children, while investing in educator training, evidence building, procurement guidance, and funding so districts can adopt responsibly.
▶ 0:25:03I'll close with a story from last year that sums up why learning about AI in schools from your teacher is so Recently, I was spending time with a friend and his daughter who's in the 10th grade. As we caught up, she'd known me from my former life as a teacher and a She found out about my new work at Magic School and excitedly said, "I use that in school." She went on to share with me that in her history class, her teacher uses the writing feedback tool with them.
▶ 0:25:33She gets personalized, rubric-aligned AI feedback on her essays that she wrote by herself, not without not with AI, from Magic School. She was grateful for the chance to get AI feedback and revise her essay before being graded by her teacher.
▶ 0:25:48Even as a high school student, she was so able so able to so clearly articulate the value of the feedback and how, without the AI tool, it would have been nearly impossible for her teacher to give meaningful formative feedback on each individual child's essay before it was turned in. Naturally, I inquired more. What do you think about AI in general? How are you and your friends using it?
▶ 0:26:13She shared an eloquent, nuanced answer, both acknowledging that some use it to but saying she's made the choice to use it sparingly. She explained that she knew that she could use it as a shortcut, like some of her friends did, but didn't want to because she didn't want to offload too much of her own She reminded me that her brain is still in a really important development phase, and she'd be tested without AI anyways, so what's the point? The conversation was inspiring.
▶ 0:26:43Not only did she clearly know the value, but just as importantly, she knew the risks and had agency to make an informed choice about how how she would use AI in and out of the classroom. The first thought I had was, "Wow, what a thoughtful, intelligent, savvy young lady." I was nowhere near as thoughtful as a 10th grader.
▶ 0:27:06The second thought was, "Wow, she must have had a great teacher that helped her learn about the risks and benefits of AI to make great choices about her own Every child deserves a teacher in a school like this. Thank you. I look forward to your Thank you. And I recognize uh Ms. Reeve Givens for your testimony. Thank you for the opportunity to testify today.
▶ 0:27:33I lead the Center for Democracy and Technology, a non-partisan organization focused on how technology impacts people's rights and futures. We have specific projects on workers' rights and the evolving use of data and technology in K through 12 classrooms, the research for which informs my testimony today. AI is already shaping how Americans learn, work, and are evaluated.
▶ 0:27:57AI can support educators and workers in meaningful ways, but without guardrails, transparency, and accountability, it can just as easily deepen inequalities and undermine progress. Building an AI-ready America means more than accelerating adoption. It means ensuring that AI lifts up all communities for sustained social and economic growth.
▶ 0:28:18Congress has an important role to play through legislation, by ensuring the proper functioning of the Department of Education, Equal Employment Opportunity Commission, and other federal agencies, and by providing the resources that workers, educators, and families need in this transformational time. Turning first to the workforce, employers are increasingly relying on AI in hiring, management, and oversight. While AI tools can create efficiencies, they can also exclude workers and violate US employment laws.
▶ 0:28:47For example, AI tools used to screen resumes have automatically rejected applicants over 55, down-rated female applicants, and filtered out resumes with employment gaps, which may be due to medical leave or returning from military service. Risks also arise from AI-powered interviews or tests, which claim to test traits like a person's appetite for risk or emotional stability.
▶ 0:29:10In one study, a researcher completed an English language AI interview in German, and yet was still determined to be a 73% match for the job. Workers often have little visibility into how these tools work or what they're actually testing.
▶ 0:29:24The harm is felt not just by workers, but by employers who are investing in ineffective tools that may break the This is why it benefits everyone to acknowledge these risks and address them through rigorous testing and AI is also being used to manage and monitor workers with impacts for their rights and safety.
▶ 0:29:44As one factory worker put it, "Any slack is perpetually being optimized out of the system and with it any opportunity to rest or recover." Workers also report losing privacy, independent judgment, and professional expertise as they become micromanaged in every facet of a task. My written testimony lists several enacted and proposed laws to address these risks, including the proposed No Robot Bosses Act and Stop Spying Bosses Act. Research shows workers strong support for these types of rules.
▶ 0:30:14Federal agencies also have an important role to play. The EEOC, Department of Justice, Department of Labor, and National Labor Relations Board previously developed valuable guidance for employers to help ensure their use of AI tools is fair and legal, but these documents have now been removed from the agency's websites. Congress should encourage these agencies role guiding employers and supporting workers at this critical time.
▶ 0:30:37In education, AI use in K-12 schools has now reached record levels with 85% of teachers and 86% of students using generative AI last school year. Each year, CDT conducts nationwide polling that asks students, parents, and teachers about their experiences with AI and how they feel about it, from which I'll share four relevant trends. First, students are turning to AI for social and emotional purposes.
▶ 0:31:02This year, 42% of students reported that they or a friend have used generative AI for mental health support and 42% reported using it as a friend or companion. This raises pressing questions about students emotional development and safety and the urgent need for privacy protections as these companies that make these tools learn so much about our and our children's lives.
▶ 0:31:24Second, deep fakes and non-consensual intimate imagery are a serious problem in schools, but less than a quarter of teachers say their schools have clear policies to respond and responses often focus solely on discipline rather than prevention and victim support. Third, more than half of special ed teachers now use AI to help develop IEPs. While most parents and students support this, it requires dedicated efforts to comply with student privacy and disability laws.
▶ 0:31:52Fourth, most public schools are now using AI-powered software to monitor students online activity, even on personal devices and outside school These tools aren't just used to support kids in distress. They're also being used for discipline. Among teachers we surveyed, 25% knew of a student who was contacted by law enforcement as a result of student monitoring.
▶ 0:32:14Five in 10 students agreed, "I don't share my true thoughts or ideas because I know what I do online is being Despite these risks, only a third of teachers have guidance on how to use these systems privately and securely. As in the workforce, AI's potential in education can only be clear-eyed about the risks and help educators and families to address them. Congress and this committee have an important role to play and I look forward to your questions. Thank you. Thank you.
▶ 0:32:43And lastly, but not least, I uh recognize Mr. Frazier for your Chairman Walberg, ranking member Scott, and distinguished members of the committee, thank you for the opportunity to testify. My name's Kevin Frazier. I'm a senior fellow at the Abundance Institute and the AI Innovation and Law Program Fellow at the University of Texas School of Law. The economic normal in the age of AI is and will be marked by flexibility. Future generations of Americans won't associate traditional work with 9-to-5 employment.
▶ 0:33:12The career ladder will be replaced by a career flywheel in which individuals succeed due to their capacity to adapt and willingness to In short, we'll soon see more and more Americans participating in what I call the portfolio economy. Workers will maintain an array of skills that they offer to a range of employers on a project-to-project basis. This transition will put pressure on New Deal era laws such as the Fair Labor Standards Act. My written testimony makes three points.
▶ 0:33:39The transition to the portfolio economy must be data-driven, worker-focused, and Absent more data about how how the economy is evolving, Congress may lack the information necessary to assess whether existing laws are functioning as intended. Many of the current sources of labor market data are infrequent, imprecise, or inaccurate.
▶ 0:34:00The contingent worker survey, for example, is issued on an irregular basis and does not accurately capture the number of Americans in non-traditional It's also unlikely that Congress has up-to-date and comprehensive information on AI adoption. According to recent polls, just 10% of workers use AI daily and a mere 9% of small firms have picked up AI. This information, though, is akin to learning that LeBron James played 47 minutes in a basketball game. It's better than nothing, but it really doesn't tell you anything.
▶ 0:34:30It's missing what really matters, like how many points he scored or, to return to AI, whether it's use is increasing productivity and altering hiring Updating and expanding the sources of information on AI progress will become all the more important as the technology continues to evolve. While most experts agree we should expect ongoing advances in AI, they diverge when it comes to pinpointing the specifics. The so-called jagged frontier of AI will cause firms to preserve the option of automating and augmenting certain tasks and roles.
▶ 0:35:00Consequently, they'll prioritize finding workers with specific skills for finite projects. In turn, workers will need to be ready and willing to learn new skills and fast. With a better understanding understanding of AI advances and adoption, I recommend Congress analyze laws such as the Fair Labor Standards Act with a focus on two aspects. First, to what extent does the law law rely on frameworks and definitions that clash with the portfolio economy?
▶ 0:35:27And second, does the law incentivize workers to engage in the career flywheel, to study, to shadow via apprenticeships, and to work in non-traditional arrangements? Thank you again for the opportunity to testify. I look forward to your questions. I thank each of the witnesses. Uh you've wetted our appetite and I guess also given us a question, were your testimonies produced by AI?
▶ 0:35:51We'll leave Under committee rule nine, we'll now question witnesses under the 5-minute I'll recognize myself uh for 5 minutes of questioning. Ms.
▶ 0:36:08Nayak, I guess the key question I would ask you is what what distinguishes high-quality uh credentials in an AI-driven market from one that is well-intentioned and Thank you, Chairman.
▶ 0:36:29What I think distinguishes high-value credentials from lower-value credentials is actually doing the work to make sure that the credentials are psychometrically valid and backed. What that means is that we're not just creating courses that people take passively and not really testing their outcomes or maybe testing their outcomes through really simple assessments, but rather we're working towards actually training AI to be able to assess people and do so against a common standard.
▶ 0:36:58Um we're partnering with organizations like ETS um to actually validate that the courses are actually leading to outcomes and we're partnering with our e-con research team that's actually following students throughout their learning journey to ensure that they're seeing the outcomes from taking the course.
▶ 0:37:15And so we think it's both a pair of making sure that the course actually has value and content that's worth learning, but then ensuring that the students that take those courses actually can apply those in their roles and then validating that in order to issue the credential rather than just having them passively observe it. So whole, well-rounded, immediate evaluation.
▶ 0:37:37Both sides of the equation. Okay. Thank you. Um Mr. Khan, there are hundreds, perhaps thousands of vendors and more will be coming all the time, I'm sure, selling AI products to schools. You mentioned that policy uh should help districts and I quote, "Ask the right questions." uh before buying AI platforms. What are what are some of those questions?
▶ 0:38:03Uh a few really important questions that a a district buyer should wonder about when bringing on an AI platform to uh their school or district is uh first and foremost, how do you protect student data? Um what data do you record?
▶ 0:38:18Are you training models with our students' inputs uh and their outputs into the other questions you might ask are how do you evaluate uh your platform for things like bias, uh safety, um ensuring that we don't see some of the the negative uh parts of generative AI uh become a part of our district. Um Magic School has put out a several white papers uh around how we approach AI safety for students.
▶ 0:38:48Uh one uh highlights the idea that AI can be uh used potentially for companionship. Um students sometimes uh will turn any tool into something that can be a companion if it's an open-ended chatbot. So what are the guardrails that you put in place to ensure that uh those kinds of things don't happen um in a public school setting?
▶ 0:39:10Um and then finally, how how do you ensure that uh the student is using it for the purpose that has been given in the classroom and not for other purposes? How do you build your guardrails into the platform so that when a student asks a question that is outside the bounds of what you want to be happening in a classroom, uh it responds appropriately? And what kind of controls do you give the teacher and the school district to uh to remedy those things and and create their own guardrails for the platform?
▶ 0:39:40Uh we know that schools and districts are different state to state Uh so and we want to make sure that there are there's an empowered district leadership to make make the right decisions for how generative AI will be used in their school or district. Okay, thank you. Mr. Fraser, when we discuss AI in the workplace, we typically put it in the two buckets.
▶ 0:39:59First, its ability to augment work performed by humans to produce higher levels of productivity and second, its ability to automate certain tasks completely and replace human effort. With the initial wave of AI in the what's been the net impact of augmenting So, thank you for that question, Chairman. I think it's important to as you did distinguish between task automation and task augmentation.
▶ 0:40:26We're seeing that even when tasks are automated, when AI is completely taking over a specific function, there are still a number of tasks that must involve some degree of human intervention. We had some people who were predicting for example the end of certain professions because they thought that AI would automate just one or two of those tasks. Those professions are still alive and well today. Radiologists, for example, are seeing higher numbers of employment, not the end of their profession as a whole.
▶ 0:40:53And so, as we look at the initial introduction of AI into the economy, number one, I'd first point out adoption is still very low. And number two, even where there is adoption, we're seeing that the bottlenecks that remain even after some tasks are automated result perhaps in even more job creation or higher wages for those workers in those industries. Okay, thank you. Thanks for answering my second question without me asking it. I appreciate that. Now I recognize the gentleman from Connecticut, Mr.
▶ 0:41:22Courtney, for his 5 minutes of questioning. Thank you, Mr. Chairman, and thank you to the witnesses for being here today. Um, you know, I want to go back to Mr. Scott's opening comments regarding the explosion of tips and calls into the National Center for Missing and Exploited Children. Um, again, the numbers bear repeating. In 2024, the number of tips that have been that came into the system was 6,800.
▶ 0:41:49In 2025, the number was 440,000 was made just in the first half of the year. And again, the the center's been around for 40 years. This committee actually authorizes its operation and funding year in and year out.
▶ 0:42:04We just actually enacted the latest update in 2024, but clearly, um, the system has just been rocked by an avalanche you know, incidents that are happening you know, this Reese Givens, I mean, you you alluded to this in in your testimony.
▶ 0:42:22I mean, what what do you what what's the response that Congress needs to make here to to hold technology companies accountable, help victims and assist these organizations to to deal with really I I think is clearly a crisis. Um, thank you for the question and for highlighting this important issue. It's this is an issue not only for young people and their parents, but it is also showing up in school communities as well.
▶ 0:42:48Our survey data this year show that 12% of students knew of non-consensual intimate imagery depicting somebody in their school community. So, that's one of their peers, it's one of their teachers. Um, and right now we really have a lack of resources for teachers knowing how to respond. Some schools are leaning into discipline. Of course, that is part of the equation, but we also need to make sure that people are talking rigorously about prevention and support for victims, including mechanisms for taking this stuff down.
▶ 0:43:17Um, there are some laws on the books that can be helpful. So, when imagery depicts minors, it is illegal content that must already be taken down. Companies can be held liable for keeping it up and for generating that content. In addition, Congress recently passed the Take It Down Act to make it easier for people to take down these images of themselves by filing a request. Just yesterday, the Senate passed the Defiance Act, which has now made its way over to the House for consideration, which makes it easier for people to sue when they are a victim of this.
▶ 0:43:48But again, putting it just on the shoulders of individual victims is not a solution. We also need companies really leaning into their obligations here. This involves a couple different things. One is about the capacity of your tool to generate these images in the first place. What are the tools in check around the creation of nude images, around the uploading of images of real people and how you allow those to be manipulated. Second is around the detection when those outputs are coming out. Why are you stopping it and how does that work?
▶ 0:44:17And then third, of course, there's social media platforms and other places where these images can be disseminated. Often times it is already a violation of the terms of service for that to happen, but we see that companies are not rigorously enforcing those terms. We advocate directly to the companies to do that. Even where it is not necessarily illegal content, but is still deeply inappropriate, they should be creating a safe community for their users and living up to the promises they're telling their users about what they do. Thank you for that answer.
▶ 0:44:46I, you know, when we enacted that recent update for the center's work, there was, you know, basically a budget sort of trajectory, which was fairly but obviously the work that they have been taking on really just within the last year is just completely been you know, we again as a society have to sort
▶ 0:45:16of figure out who pays in terms of dealing with that, but you know, this is a completely consensus issue in terms of the center's authorities. It goes back 40 years, I think as I mentioned earlier, and it should be frankly, in my opinion, a consensus issue for industry as well as the political parties here in Washington.
▶ 0:45:37So, again, at some point it does come down comes down to the question of who pays and and it seems that industry has an obligation here given the fact that there's a direct connection to to, you know, the scaling rapidly, I think was the what the chairman said about AI at the opening and and really what's also rapidly happening in terms of negative Uh, no, sorry, I very much agree and I think the question here is both the technological developments, so what are those
▶ 0:46:07safeguards that should be embedded into the tools, and then the will to actually follow through. It's one thing to write a good terms of service saying that you don't allow this on your platform, it's another to rigorously enforce it. One of the problems we write have right now is lack of visibility into how platforms are complying with their terms of service. Some of them do do disclosures about it, but it's not a common practice and we've actually seen some pullback in recent years.
▶ 0:46:31And so, when we think about accountability, when we think what it is to make good on the promise you make to your customers and your users about the community that you are building online, about the value of these tools, how can we get far more information and honest disclosures around how these issues are being addressed in a rigorous way? Thank you. I yield back. I thank the gentleman. I now recognize the esteemed gentleman from South Thank you very Columbia, South Carolina. Lexington. Lexington.
▶ 0:46:57Anyway, the bottom line, I'm just grateful to be here, but I thank you, Chairman Tim Walberg, and it's really great to see the bipartisanship with Ranking Member Bobby Scott, Joe Courtney, the whole team. And it's because what we're dealing with is just so important and then thank all of you for your testimony today on such an important issue. And Mr.
▶ 0:47:20Khan, with this in mind, how and you've hit on this already, but how can we promote that the presence of AI in K through 12 education is not diminishing a student's ability to think particularly children who are not aware of the risk. And additionally, how can we ensure the protection of minor students' data when they utilize AI? Thank you for the question. I think that there are a couple of things that we we believe as a organization.
▶ 0:47:49One is that if a student is not introduced to generative AI in the school building, they will inevitably be introduced to generative AI in a different way. So, that might be that, you know, you open Snapchat and you see My AI at the top of it. You might open Instagram and see a an AI chatbot in your search function.
▶ 0:48:11I think that everyone in this room would agree that you'd prefer the first time that a child interacts with generative AI, it's under the guidance of a teacher who's teaching them about how it works, the risks and benefits of using generative AI, the purposes with which they could find value and also the risks associated with the use of the technology, one of which I didn't mention in my testimony, which is offloading too much of your own thinking. We don't want students in primary school having AI write for them.
▶ 0:48:40We want [clears throat] students having the writing done by themselves. I was an English teacher and I think so much about how, you know, writing is beyond the finished product. It's the construction of thought, right? It's you are developing your own ideas, your own opinions when you put your pencil to paper.
▶ 0:48:57So, we want students doing that and we don't want them offloading that to AI, but that starts with teaching students about the technology and the risks and benefits of it and actually engaging with it to understand how they can be productive users of it in the now and in the future. I appreciate your thoughtfulness and Ms.
▶ 0:49:17Dyak, with about a third of Americans reporting that they're using chat developments like OpenAI's jobs platform has immense potential to affect the community. Could you elaborate on how you see this job platform working and what outcomes you're hoping to see. Thank you, Congressman. Yes, our goal is to certify 10 million Americans by 2030.
▶ 0:49:45And with that certification, what we're trying to do is build a jobs platform that can then connect those Americans to roles across the economy. And so our goal here is first when we launch, we'll start with a product that's focused on users of ChatGPT and helping them get career advice, helping them understand where their skills are, what gaps they may have, helping them actually get AI skills if they so choose through our certifications effort, and then finally helping them connect with employers that really value those skills.
▶ 0:50:15Certifications is really the through-line for this whole effort because we believe that we can actually have employers trust those certifications and then be able to use them to help them identify roles for different people in the economy. Well, again, all of you, your expertise is going to be so helpful. And Professor Fraser, you wrote in a recent op-ed, quote, "The AI debate is being framed by questions that have distracted us from the practical policy challenges we need to address to bring about a better future." End quote.
▶ 0:50:46Will you expand on that point and explain how you think that we should be talking about AI's potential impact on the workplace? And I I have a practical problem that I've seen in that I have found out that a number of young people don't know how to read cursive. And I'm concerned as they're so AI focused that um I'm having difficulty communicating with people who can't read cursive.
▶ 0:51:11With respect to cursive, sir, I think I'll leave that to the English professor about the the merits of writing in that style. I will instead focus on the question of how do we make sure that Congress prioritizes the issues that really matter in this AI debate. I think to my colleague's point, making sure we focus on AI adoption and AI literacy is incredibly important.
▶ 0:51:34Right now many people have access to these tools and our polls that we issue ask them, "Do you use, for example, ChatGPT?" Well, sure, but you may be creating a meme. That's like asking someone if they've ever driven and they say, "Yes, Mario Kart all the time." Right? It's not a meaningful metric of whether you actually know how to use the tool. And so we need to have a far greater focus on making sure we're paying attention to literacy and true skill development.
▶ 0:51:58Well, I wish you well, and I've seen the good and the bad. I a political candidate actually made fun of me doing the Macarena, which I don't know how to do, but it made me look good. And then and then I don't think they meant for it. And then recently I've got a beautiful about a a showing me showing a picture of a beautiful lake and showed that I was not there. But I what a interesting time we all live in. Thank you, Mr. Chairman.
▶ 0:52:30I thank the gentleman for his Macarena. And now I want to recognize the gentlelady from Oregon, Ms. Bonamici. Thank you, Mr. Chairman, and thank you to the witnesses. Special welcome to Professor Fraser, who is in Texas now, but via Oregon, an Oregonian. So so we know that AI is rapidly reshaping classrooms and workplaces. I'm very concerned about the minimal if any professional development on how to responsibly use AI in classrooms. Mr.
▶ 0:52:58Khan, thank you for your mention of guardrails and the human elements, and I appreciated your story, but the last time I spoke with high school students, they told me they're terrified that their classmates are going to use their images in a way that is violating their rights. So so I'm a member of this committee, education workforce. I'm a member of the Committee on Science, Space, and Technology, where I'll note there's also an AI hearing happening right now.
▶ 0:53:21And I also served as bipartisan member of the AI task force. And I have heard and raised questions for years about security, bias, the ethical use of AI, and also the challenges of enforcing violations of civil rights and privacy rights in the context of of the use of AI.
▶ 0:53:37It's important to note that AI literacy isn't just about how learning to use the tools, but also how to critically evaluate them, to critically evaluate AI through the lens of ethics, humanities, linguistics, and other fields and perspectives that are often underrepresented in these conversations. So Ms. Reeve Givens, first thank you for mentioning my bipartisan No Robot Bosses Act, which we have reintroduced in this Congress. Thank you to Representative Moilanen this committee for co-sponsoring that.
▶ 0:54:06It's HR 6371, so I invite all my colleagues to sign on to require employers who use automated decision systems to provide specific disclosures. But my question today is about the need for a broader perspective in AI and what we need both in the short term and in the long term. I submit that it's not just technology skills, as I mentioned.
▶ 0:54:30We need people with backgrounds and fluency in ethics and history, linguistics, humanities, and other fields that develop as Representative Wilson was saying, critical thinking skills. So how will these perspectives [laughter] and backgrounds help to equip students and workers and employees employers to better develop, but also responsibly use AI and navigate its challenges, but also its opportunities? Thank you so much for that question and for your leadership on these issues, and I think the way that you frame it is exactly right.
▶ 0:54:58To make good on the promise of this technology, we have to have a long view on what some of the both what some of the immediate and long-term risks can be, and have that grounded in the people with appropriate subject matter expertise. The subject matter matter expertise is not just AI, it is learning science, it is student emotional well-being, it is privacy and And so and then in different domains, when workers are now having automated management systems, those workers also have domain expertise where the AI can be used
▶ 0:55:28to enhance their skills, but they need a voice in how it is being deployed for them to really make good on that specialization and that promise. Just to reinforce some of the concerns that you were raising about the use of AI in schools, I will say that our survey data very much reinforces what you are saying, which is that you know, 72% of parents, 71% of teachers do worry that AI will diminish students' critical thinking skills. They are really thinking actively about what training is needed for teachers, and teachers are asking for more.
▶ 0:55:56Of course, it is about how to use these tools and how it can help them in the classroom, but it is also necessary to underscore what the potential risks can be and how teachers can play a role in mitigating those harms. You you actually anticipated my next question, which was about how the lack of educator professional development, which you established, I mean I've read the the statistics in your testimony. It's it's it's pretty concerning.
▶ 0:56:21How does that increase the risk of students not only improperly using AI, but experiencing privacy and security security
▶ 0:56:29Well, in two different ways. I First of all, Mr. Khan put it perfectly when he said that teachers really can play an important role helping steward students through this transition, helping them understand how to use these tools, how to think critically about them, but the teachers need to be prepared to do that. One of the I think the things that's important to emphasize here is right now just the lack of transparency in a lot of the ed tech products around how they address some of these concerns. Mr. Khan also spelled out what some of those considerations can and should be. Are these tools grounded in learning science? Have they been tested for bias?
▶ 0:56:59Are they Do they have appropriate guardrails for use by young people? What are their security and privacy protections? But right now we don't have a standardized rubric. I will say that CDT has published one. We have a rubric of eight factors and have assessed 100 ed tech vendors against it, and we're still just not seeing disclosures around this. And that's stopping educators both at the teacher level, but also at the school administrator level, making informed decisions about what tools to use and have a much more evidence-based conversation around this progress.
▶ 0:57:28I appreciate that very much, but I also note that this administration is trying to shut down the Department of Education and eliminate or extremely cut funding for things like Title II, which is professional development for educational research. That's We need that more than ever, especially in the age of AI, because we know we can't achieve an AI-ready workforce without a national strategy, without a dedicated investments, and we can't lead the world in AI by leaving students and educators and workers as an afterthought.
▶ 0:57:57As I yield back, Mr. Chairman, I'm developing a comprehensive legislative framework to strengthen the teaching, learning, and dignity of work in an AI-enabled economy, and I look forward to working with my colleagues. I hope on both sides of the aisle to improve AI literacy and workforce readiness while protecting privacy and security rights. And I yield back. Thank you, Mr. Chairman. I thank the gentlelady. I recognize the chairwoman emeritus of this committee, Ms. Fox from North Carolina. Thank you very much, Mr.
▶ 0:58:27Chairman, and I want to thank our witnesses for being here today. This really is an important topic that we all need to be concerned about. Ms. Nayak, based on your work leading certifications and the jobs platform of OpenAI, what is one major misconception policy makers often have about how AI will transform credentialing and hiring? And why does that misconception matter for workforce and education policy decisions?
▶ 0:59:01Thank you, Congresswoman. I think one major misconception is the speed at which jobs will be impacted by AI. What we're really seeing right now in terms of the trends that we're looking at is rather than jobs being lost immediately, rather jobs are being augmented. And we're seeing people start to introduce AI tools into their everyday workflows, like getting writing advice, guidance, rather than kind of taking over their whole roles.
▶ 0:59:29What that means for certifications and our jobs platform is now is the time to give AI workers skills and we think we can do that through our certifications effort. And then if we're able to build um on the fact that people are already coming to AI to learn, we can certify them to learn, then we can actually work with the economy, work with employers to actually then connect them to opportunities that leverage those skills and I think that will help us transition um and I think it will be a lot less of just
▶ 0:59:59this kind of rip of the band-aid where people, you know, lose their jobs Uh it's unfortunate we were not able to pass a bill yesterday with give employee employees the opportunity to work more on their skill development using AI um without it um impinging upon their I have another question for you.
▶ 1:00:23There are many conversations happening around automation versus augmentation in an AI-enabled workplace. Could you speak on what you believe will be the durable skills workers need to develop to succeed in an AI future and how AI technology can help develop those Yes.
▶ 1:00:46So, um one of the things that we're doing with our certifications effort is we're building a series of courses called AI Foundations and we think those will be the foundational skills that people need to be able to effectively use AI. Our coursework starts with AI Foundations. Um this one's focused on chat GPT, which is how do you actually prompt engineer a model, how do you work with the model to co-develop outcomes. Our second course will focus on applied AI skills.
▶ 1:01:13How do you actually use models in um being able to create agentic workflows. Um I spent a lot of time with frontline workers and that's actually the course they're most excited for because they see it as being able to take on parts of their job that they really don't like doing.
▶ 1:01:28I spoke to people unloading trucks um at kind of large um department stores and they really want to automate the workflow of choosing what truck um to to kind of offload first and rather than having to, you know, decide themselves and then make the wrong decision and cost the company money. The third course will focus on how do you actually teach others around you AI and I think this speaks a lot to um what Mr. Khan was saying about um leveraging AI in the classroom and being able to enable people to train others.
▶ 1:01:58Um we're looking at a AI leadership and training course where people actually learn skills like responsibility in AI and how do you effectively train others around you to co-leverage AI with you. Um so, we think those are the three foundational courses and then um our hope is to actually build those out to more specialized courses for industries. Great. Thank you very much. Um Mr.
▶ 1:02:20Frazier, your written testimony points to several sources of information for data on AI's impact on the workforce, but you say these sources point only to paint only some of the picture. Should robust information gathering and information sharing on AI's impact begin in the private or public sector? So, thank you very much for that question, Congresswoman. I think that we need to put a all hands on deck mentality approach to data gathering at this point.
▶ 1:02:50Uh if you talk to any good economist, they'll tell you first they don't know what's going on in the economy because we don't have enough information about who's adopting which tools and to what ends. And so, having updates to our information gathering techniques starting with, for example, the contingent worker survey at the Bureau of Labor Statistics. This is issued on an ad hoc basis. Uh there was a gap from 2017 to 2023, for example. We know that the gig economy in that window transformed tremendously.
▶ 1:03:19So, let's start both with issuing more frequent calls for information and also making it easier for companies, private sector entities, to submit and share information. I believe that AI will facilitate that process and I believe that we can make sure that we aggregate and anonymize that data such that compliance rates are pretty high. I have a short follow-up and I'll ask you to just submit the information later. Please let us know if there ways we can encourage information sharing that'll benefit both sectors.
▶ 1:03:48Thank you, Mr. Chairman. I yield back.
▶ 1:03:51I thank the gentlelady and I recognize the gentleman from California, Mr. Takano. Thank you, Mr. Chairman. Uh thank you to the witnesses for being here. Mr. Frazier, I would like to begin with you. In your testimony, you predict that AI will generally accelerate workforce trends that were {quote} already underway before chat GPT {end quote}. You predict that the future of work will look far more like a gig economy than a multi-decade career at a single firm.
▶ 1:04:19Can you expand on the workforce trends you describe that will be accelerated by AI? Yes, thank you very much for that question, Congressman. Uh I want to start first with the trend that we're seeing called fissuring. Uh fissuring was coined by David Weil, who's a member of the Department of Labor last administration. And this is the idea that we're seeing more and more lead companies, think massive Fortune 500 company, segmenting off certain tasks and skills to what he labels subsidiary organizations.
▶ 1:04:47So, that janitorial service that used to work at Alphabet, for example, is now purely done on a contractual basis, not housed within Alphabet itself. As we see this fissuring of the economy, the result is that these subsidiary organizations are held to high standards of efficiency and effectiveness, which means they need to find labors who can come help out on a project-to-project or short-term basis and fulfill that very specific niche uh skill portfolio.
▶ 1:05:16So, as workers adjust to this fissuring concept and as they adjust to the fractionalization of work generally, they'll need to maintain a portfolio of skills so that when the next opportunity pops up or that new project arises, they're able to find a uh organization that wants to hire them for that purpose.
▶ 1:05:34And this is what you mean by the gigification of work, that there's a series of gigs. Uh I'm not sure if I used that term, but if I did, yes. Um so, the gigification will result uh more or less in people not working for a single entity for multiple years. Uh I will point out we're already seeing this, for example, among Gen Z. The average job tenure of a member of Gen Z within the first five years of their career is 1.1 years. They're not sticking around for a very long time.
▶ 1:06:02Well, so to clarify, you're saying that with the with the integration of AI into workplaces, American workers will have much more varied employment structures and arrangements and less of the traditional 9-5. Is that correct?
▶ 1:06:12Very much correct. I think we already see, for example, that there are 71 million workers who have opted into uh independent work. Okay. Well, given those rapid changes to the labor market and whether Americans thrive or not will depend on whether labor laws and protections evolve alongside the economy.
▶ 1:06:29Uh this graph that I have behind me uh shows the trends of in the US labor productivity uh about labor productivity over the last 50 years with this orange line here you know, showing uh that productivity. It's it's a shows an increase. Uh Mr. Frazier, what would you project uh the that orange line that orange productivity line to look like over the next 50 years if we continue on our current trajectory with the implementation of of of of AI?
▶ 1:06:59I would project that we continue to see increases in labor productivity over time. As we've noted previously, economists are looking at the integration of AI into the workplace and we're seeing that in many regards it's augmenting work, increasing productivity because AI is not at a point where it's automating every single task. Instead, it's leaving a number of tasks that require human intervention, making those tasks more valuable and more remunerative. Well, so we can expect that productivity to increase potential Some would say even exponentially.
▶ 1:07:29Um that the line that I'm concerned with, however, the most is the blue dotted line uh which which represents real hourly compensation. As you can see, workers' wages have not kept up with the increases in their productivity. In fact, they've increased only about 30% in 50 years, while productivity has increased by nearly 90%. This huge disparity has been a major factor in the increase of wealth inequality in the United States.
▶ 1:07:54And as we speed toward a future in which AI is a standard feature of the workplace, we must ensure that the benefits of this increased productivity are reaped by the workers, not just tech executives and the shareholder class. Although AI represents uh presents an enormous opportunity, the real risks of worker displacement and increased wealth inequality must be addressed head-on. Mr. Chair, I ask unanimous consent to enter this letter from the uh from AFSCME on behalf of 1.4 million uh of their members underscoring this point. Without objection.
▶ 1:08:25Now, one clear solution uh is through the strengthening of labor protections, a higher minimum wage, increased overtime threshold, protected uh and protect and protected right to organize uh will be uh the will be key to ensuring that AI benefits all of us rather than than making this gap um this gap here larger. Uh my 32-hour workweek proposal, the my 32-hour workweek bill, addresses the need for greater worker protections in the new economy, uh the new American economy.
▶ 1:08:55My bill, which will be reintroduced this year, would amend the Fair Labor Standards Act to make the standard workweek 32 hours rather than the 40 hours with no reduction in pay. If the future of work is one with much higher output per worker, varied employment arrangements, and a trend toward a gig economy, then our baseline labor laws must adapt as well.
▶ 1:09:18AI presents an enormous opportunity for American workers, but AI should be seen as a tool uh that works for workers and not again instead of against them. Thank you and I yield back. I thank the gentleman. Just make sure I took notes correctly, Mr. Fishering or fittering? Fishering. Fishering. Fishering, yes. I I'm still hurting from the Michigan's loss to your school, so I wasn't
▶ 1:09:48Blame blame Arch Manning on that one. Not not my fault. Okay. So, Fishering. Thank you. Now, I recognize the gentleman from uh Wisconsin, Mr. Grothman. Yeah, Ms. Nayak, uh you talked a little bit about How do you view a credential or um what does that mean?
▶ 1:10:08I mean, if I'm looking to hire somebody and I want to know if they're familiar um with with this new certifications. Who Who determines this? So, thank you, Congressman. Uh for the credentials that we're building here at OpenAI, um what we're hoping to do is create several credentials, um an AI Foundations credential, and then credentials that are focused on industries. And who runs these things?
▶ 1:10:38Um OpenAI is running it, but with a partnership of employers, um NGOs, as well as universities, um and we're kind of partnering across the coalition to build the credential.
▶ 1:10:49how do you invisionize that I succeed in that credential? Do I have to take a class? Do Does my employer just say, "Glenn knows how to do this. He does it very well. We think he's okay." I mean, credentialism is always a potential problem, I think. I I don't like it, but I just wondered 5 years from now, how would you get an AI credential? Going to college, you know, um being trained at work. How do you invisionize that?
▶ 1:11:19Yeah, thank you, Congressman. The The way that we think about credentials is we have a course that's structured, and it's actually accessible on ChatGPT. So, ChatGPT is actually taking that curriculum, personalizing it to the user, understanding their needs, and then it's um actually judging them on their ability to retain the material, similar to how a teacher would in a class. Um the difference is that it's immensely scalable because anyone can access it, um and uh you know, it's personalized to them.
▶ 1:11:48And and therefore, it's something that is unrelated to say what we have a formal college degree. It could actually be related. So, a lot of credentials
▶ 1:11:59have to be. I'll put it that way. It wouldn't have to be.
▶ 1:12:01Correct. It wouldn't have to be. I'll ask I'll ask the other couple. You We'll move over one. Do you think we need more credentialism in this field? And And I'll put it this way. When you see people being hired, Mhm. um is the employer looking for some sort of credential or is it just I'm looking at your resume and you say you're Mr. an expert in AI, that's good.
▶ 1:12:30You You must be doing it in your current Right now, how how is an AI skill being determined? Yeah, so internally I We're a fast-growing company and we're hiring folks, and we certainly value uh certain skills uh like potential potentially the use of generative AI. One component of all of our interviews and our processes is doing on-the-job work as a part of the process.
▶ 1:12:58We pay uh prospective employees for that work. Um and we don't prohibit them from using generative AI in that work, but rather than uh saying you can or cannot use it, we're saying, "What's the quality of the final outcome?" If you use generative AI a lot and the final outcome of that product is really high quality, fantastic. If you didn't use it at all and it's super high quality, fantastic. Um but the The focus is on the quality of the work rather than whether or not you use generative AI. Okay.
▶ 1:13:28So, you have There's no credential right now, and it it doesn't affect how your business operates. Mr. Frazier, same kind of same questions down that line. Yeah, so something I think that's important to add to this conversation, Congressman, is the importance of thinking about alternative ways for uh firms to experience bringing on a new employee, for example, through apprenticeships. Right now, we have about 600,000 apprenticeships across the US.
▶ 1:13:53I think we could drastically increase that number to make sure that folks have an on-ramp to demonstrating and showing their capacity of certain skills. I'd also argue that in this context, we always have to be asking, "Compared to what?" Right now, our main credentialing approach is through four-year universities. Um I'm not going to talk about my own uh curve that I have to apply to my tests, but I'll tell you that at other universities, you will find uh at the end of the term, a professor has one of two options, A or B.
▶ 1:14:23And so, when everyone has a 4.0, uh it's very difficult to assess if that 200 grand in higher education was worth the signaling cost. Okay. Now, I'll give you another question. On AI, as we generate things, uh there can be a viewpoint with what is generated. It could be affect how one looks at religion, it could affect how someone looks at politics, it could affect how someone looks at history.
▶ 1:14:49And one of the things that scares me as people get a hold of that uh you'll affect or you'll push a politically correct view on all these topics. Is this a concern and how do we deal with it? Uh briefly, I'll say this is very much a concern. I think that how you make a cake is the same way you make AI. It involves a lot of ingredients and it involves a lot of taste, and the different tastes of certain individuals will manifest in that outcome.
▶ 1:15:18The best way forward, though, isn't to impose certain top-down views on what models have to do or what they must be trained by, but in my opinion, fostering competition in the same way you walk down a movie aisle and you can find a movie for every taste or for every genre. We should make it so that American consumers have access to a wide range of tools with a wide range of perspectives. The gentleman's time has expired. I now recognize the gentlelady from Georgia, Ms. McBath. Thank you, Mr. Chair, and thank you to our witnesses for being here today.
▶ 1:15:48Uh there are a few issues that stand to be as impactful to our society as AI projects are to be in the coming
▶ 1:15:55Oh, no luck. From our schools to our job sites and hospital systems and everything in between, AI is already changing not just the activities that people do in these places, but how we do them. We are living through another industrial revolution, so to speak, and Congress must provide clarity, which I hope hope we are able to do today, for states and ensure that Americans understand how to make I AI work best for them.
▶ 1:16:25In Gwinnett County in Georgia, where I I reside, the largest school district in kindergartners at some schools have already begun to use AI uh combined with LEGOs in classroom to solve some problems. At 5 and 6 years old, uh they may not understand what AI stands for just yet, but they know how to use it already. And they understand that it will also be a big part of their future.
▶ 1:16:55Programs like Gwinnett's Computer Science for All ensure that children are made comfortable with using AI at a year a very young age and that they understand how important it is. Yet, these applications can also be used to make funny videos, but it is so important that we cut out the noise for our students and help them understand how our world is changing and what it really means for them.
▶ 1:17:23AI has incredible potential to be a positive force in society, but that doesn't happen on its own. Congress must ensure that people who lose work and possibly their only source of they cannot be left behind. We've seen this before. We cannot allow what happened to the factories and the mills across this country to happen to other parts of our economy as well.
▶ 1:17:52When these jobs die, the communities that rely on them, they die a slow death as well. Drawn out deaths alongside them. The disappearance of good-paying jobs that didn't require people to go into huge amounts of debt to get them is why the American worker makes less today than they did in the 1990s. It's why there's so much frustration in this country about the cost of living.
▶ 1:18:22About how difficult it has become just to afford the things that they used to used to be commonplace for middle and working families across this country. This can't be the norm when it comes to automation and dare I say innovation. It has only become more expensive for people in this country to get the skills necessary to afford a decent life.
▶ 1:18:46And I'm deeply concerned that if they're not implemented correctly, AI might not achieve its full potential to make make life better for people. Instead, it could make it even more difficult for people just to provide for themselves or just to put food on the There are millions of Americans who've done everything right and have nothing in this country to show for it.
▶ 1:19:09Despite going to school, paying for a degree of some kind, or learning directly on the job, so many Americans find themselves no closer to the financial freedom and independence that I put all of their effort towards to begin with. With that, comes feelings of shame, feelings of resentment towards only our not only yourself, but the system that you put faith in. Ms.
▶ 1:19:34Reeve Givens, what sectors of our economy and and occupations are the most at risk for displacement and can you give your best estimate for when we might begin to see major disruptions or significant drops in the numbers of jobs that going that are going to be available for real Thank you for your question and I think one of the things that it ties together nicely from every witness down this day as I think we're hearing that improved training in AI skills
▶ 1:20:04really matters, but it is only one piece of the pie and if Congress only focuses on that piece, they're overlooking ways that we really can have an AI-ready economy and society. To answer your question specifically on worker displacement, this is less the focus of my organization, but of course we follow the research closely and a lot of the research does focus on tasks such as in the office and administrative occupation And so one of the unusual turns that we're seeing is that this is less about the changing of jobs in the kind of the factory
▶ 1:20:35worker space, although I think that may come as well, but most immediately with generative AI, we're seeing displacement of tasks um for things that do relate to kind of office administration, etc. Research from the Brookings report says that 30% of all workers could see as many as 50% of their occupational tasks uh disrupted over time and while that doesn't necessarily mean full displacement and as as Professor Fraser said, we're still seeing how the evidence bears out, this does does point to significant changes including in this
▶ 1:21:05class um of administrative occupations which have been the pathway to upward mobility for many people across our society including those without college degrees and overwhelmingly women as well. So it is a new group of people for whom we have to think about the significant transformational change. Um those also tend to be communities that are not organized by as a union or have representation and so again, we need to think about how we protect those rights. Thank you very much and I am out of time.
▶ 1:21:34The gentlelady's time has expired. I now recognize the gentleman from the Masters Golf and chair of the HELP subcommittee Mr. Allen. Uh thank you, Chairman Walberg for uh holding uh this hearing and I want to thank our witnesses for your expertise on this important uh topic today. Uh Ms.
▶ 1:22:01uh Nayak from AI's perspective and we're talking about worker skills credentials here. What would it take for these skills and credentials to be more transferable across jobs, industries, or regions? Thank you, Congressman.
▶ 1:22:22What we think will make these credentials really transferable and and certainly the way that we're building them is we're building a foundational set of courses um that we're calling AI Foundations, but then we're making it so that they can be adapted by industry. And so as an example, um we have our AI Foundations course, but we're partnering with employers like Walmart and Lowe's to think about how can we adapt that course to frontline workers? How can we adapt it to administrative workers and office workers?
▶ 1:22:50How can we adapt it to AI builders who are actually using AI to build tools for those workforces? And so um the way that we approach it is you build the foundation, but then with AI, you can truly make those courses adaptable to different industries. Uh Mr. Khan, I've long believed in the importance of the family and education. Parents should make the final decision in their child's education and responsible parents are key to making sure students get the education uh their children get the education they need.
▶ 1:23:20That principle is especially important when it comes to to to AI because these tools can often be accessed in the home. What advice would you give to parents who want to make sure their students are using AI responsibly? Uh thank you for the question, For families, I think it's really important that they are having the conversation at the kitchen table.
▶ 1:23:44Uh you know, what are the ways that you're seeing generative AI affect your life as a child in school? Some of the dangers that we've heard of generative AI today uh in the hearing are things that should be brought up and questioned and understood.
▶ 1:24:00Uh I think that uh having generative AI be uh a topic of conversation and uh an education that is granted not only for teachers, but also for families, literacy for generative AI is incredibly important for uh for schools and teachers, those who are, you know, imparting this wisdom upon kids about how uh this technology works and the risks and benefits.
▶ 1:24:24Uh but you know, second to the teacher in the school, the family also is a really important part of a child's uh success in the future. Uh I'll speak to a just I'll try to make this as quick as possible with a story from when I was a teacher. Uh Uh this was right when social media was taking off. We were seeing cyberbullying uh rampant in our schools and families didn't know what to do. Their kids were coming home on Instagram or or Snapchat?
▶ 1:24:53And uh I actually held at our school building a uh an information session for families to teach them about the social media tools that their child their children were using that they were just happening under the feet of parents. They didn't even know that these uh this is before Instagram had really taken off. Uh and you know, helping parents become informed guardians of how they choose their own children's use or not use of technology.
▶ 1:25:17My recommendation was you probably shouldn't have your middle school child using these tech these uh these apps, but you know, with when you don't even know what those apps are, it becomes really really challenging. Uh the school building ends up again being a really powerful place for families to learn about this technology and to make their own informed choices about how their kids will or will not engage with technology. You know, so many years later, we're seeing cell phone bans in schools. We knew this 20 years ago. We knew this shouldn't have been something that was a part of the school environment.
▶ 1:25:47We can't repeat the same mistakes of the past with generative AI. We need to get families informed, schools informed, teachers informed um and we can write a different uh script for the the future of how this technology plays out for our children. That's great. Well, thank you for sharing that that story. Uh Mr.
▶ 1:26:04Fraser, uh according to a poll published in October, 40% of American voters cite cost of living concerns as their number one issue in the upcoming midterm You have made the case that AI will make life uh more affordable for everyday How do you believe this technology addresses affordability concerns and can you give some current examples
▶ 1:26:25uh or or examples that come will come in the near future? We got about 22 seconds, sir. To describe the full benefits of AI in terms of affordability in 20 seconds is quite the tall order. I'll direct you to uh a op-ed from me and Adam Thierer in The Hill on affordability in AI. For right now, I'll just bring one very short story. Uh the University of Texas at Houston health system used AI to predict when folks were going to miss their They saved $250,000 in one quarter.
▶ 1:26:55Many Americans can't afford healthcare. We should be exploring for example in healthcare how we can automate and improve uh describing and so many activities that drill uh increase our medical costs. That's a great example and be feel free to submit more in writing. Uh it would that would benefit the commit committee and Mr. Chairman, I apologize for running over. I yield back. I thank the gentleman. I recognize the gentlelady from Pennsylvania, Ms. Lee. Thank you, Mr. Chair.
▶ 1:27:24I represent Pittsburgh and Western Pennsylvania, a region that reflects both past innovation uh and manufacturing and steel and of course, future uh advancement in technology as an emerging artificial intelligence hub. So my district embodies the evolution of industry and adapting to new technologies.
▶ 1:27:42And also represent communities that were left behind um by innovations in industry, communities that are still reeling from harms uh and the legacy of redlining and and So I'm here today to urge us to not repeat um some of the errors of the past. AI's algorithms are trained with human data, we we we know and we can recognize that.
▶ 1:28:08So they learn every bias and every uh inequity that is already entrenched in our society and it is the uh of this body to ensure that AI is not weaponized as the next tool to streamline systemic uh discrimination.
▶ 1:28:25Republicans love to talk about the cost of regulation that businesses will be bogged down when spending on compliance standards uh that will hinder innovation, but what about the costs to say a child who is traumatized by a police encounter because their clarinet or a bag of chips was mistaken for a gun by their AI's weapon the school's AI weapon detection system or the child that will have AI-generated images of themselves uh living on into perpetuity on the internet that folks would be able to see for
▶ 1:28:55the rest of that child's life. Or what about the person who's misidentified and then docs because a facial recognition system can't tell black people apart. What about a person who doesn't get a job because of the AI screening tool uh being used discriminates against their application? How do we quantify those costs? Nearly uh 90% of companies are using some form of AI for initial job candidate screening.
▶ 1:29:18So, Miss Reed Gibbons, are these companies required by federal law to conduct ongoing audits of algorithmic bias or testing for compliance with for civil rights law? Not yet. Um, but they are liable if they are discriminating against candidates.
▶ 1:29:34And so, this is why it's in the interest of employers as well as workers for us to have a far more rigorous approach to auditing requirements, to transparency requirements, so that employers, too, aren't buying snake oil products or products that are inadvertently or intentionally discriminating against people that are applying for jobs.
▶ 1:29:49Definitely, and I appreciate that, and I agree. But, for those who aren't you how are we to be sure that the tools that they're using aren't exacerbating systemic inequities in hiring? So, there are a couple different pieces to this story, and one is the urgency with which employers need to understand what is at stake when they are using these tools, the need for them to use them responsibly, and not just say that as a hand-wavy statement, but genuinely take concrete actions.
▶ 1:30:16My organization and others have set out a good list of best practices, the EEOC, the Department of Labor have also put out similar guidance for employers on how to be compliant with employment laws and not discriminate. And the things that they require are things like notice to workers, so you know how you're being tested, and knowing how you're being evaluated. Have you done appropriate auditing? Are you checking the system after it is being used in the workplace? All of these are very common sense things. It's not burdensome red tape.
▶ 1:30:41It is simple things to make sure that these tools actually work in the way that they are intended, and all workers can be considered equally. Thank you. And if I could add to that, the AI Civil Rights Act, the bill I introduced with Congresswoman Yvette Clark and Senator Ed Markey, and then the Eliminating Bias in Algorithmic Systems Act is a bill that I'm going to be introducing this week with Senator Markey again. It will put real guardrails in place that demand action towards transparency and accountability and strong civil rights protections, because we we can't afford to do nothing here. Mr.
▶ 1:31:10Khan, last summer Common Sense Media identified racial bias in Magic School's AI-generated behavior intervention plans. In light of these findings, of course, you all said that you were going to work to improve your bias detection systems. Yet, your website still lists both behavior intervention suggestions and behavior intervention generator under all teacher tools. If I could ask, why hasn't Magic School removed the behavior intervention tools as you work to ensure they're not racially biased? Uh, thank you for the question, Miss Lee.
▶ 1:31:38I was actually in Pittsburgh recently meeting with educators who love Magic School. Um, uh, one, uh, we attempted to replicate the findings in that study around racial bias in those tools. We were unable to replicate those findings, but we did, uh, request, request from uh, Common Sense their data. Um, we still do not, uh, we still stand behind our tool. We don't believe that there is racial bias in our tools, uh, but we do receive feedback. We serve schools and districts all over the country.
▶ 1:32:08I've worked in schools and districts all over the country, and we value equity, um, and, uh, strive to remove bias in any of our tools. Thank you. I I I can appreciate that, and I just hope for the obviously the sake for the students that that is true, because what was found there could be, uh, concerning, but it also replicates some of the the findings that we had just in everyday interactions with students. That students who are black or brown or poor or students with disability are more likely to be punished more harshly.
▶ 1:32:36so, I wouldn't be shocked, but I I appreciate you all looking into that with urgency, and also, of course, trying to innovate in ways that will be helpful in the future. It's important that we continue this work. It's important that we do it in responsible ways. Uh, with that, I yield back. I thank the gentlelady, and I recognize the vice chairman of the committee as well as the, uh, subcommittee chair for higher education workforce development. Very interested in this topic, uh, Mr. Thank you, [clears throat] Mr. Chair.
▶ 1:33:02And, uh, first of all, appreciate this very timely, um, hearing as we embark upon one of the more exciting, uh, times in our history, in our country's history.
▶ 1:33:13Um, before I get started, I just wanted to, um, uh, as we talk about preparing for our much-needed and robust workforce, to highlight that the Department of Education, uh, launched the Connecting Talents for Opportunity Challenge It marks a crucial first step in modernizing our workforce by connecting the worker skills and credentials to education and job opportunities. I want to thank, uh, President Trump and and Secretary of Man for the support on this initiative, and look forward to, uh, to seeing this further develop.
▶ 1:33:41Uh, Miss Naya, uh, from from your work on on certification of job platforms, how important is interoperable learning employment records uh, for making AI, uh, enabled hiring actually work at Thank you, Congressman. Um, at Open AI, we think interoperability is essential, um, for our credential work as well as for our jobs platform, because it will mean that people can leverage multiple platforms, they can get educated, get skilled, but they can take it with them.
▶ 1:34:11Um, one of the things that we're doing for our certifications platform is we are connecting to, um, technologies that allow people to take their credentials with them, and then they can use those, um, across like their resumes, across their portfolios, um, with employers, and employers will be able to also validate those skills. Um, so, we believe that interoperability is quite important.
▶ 1:34:32I I think that term should be, uh, highlighted as something that the American people need to be educated on, cuz we we hear that, we talk about it, um, but we have no idea that the reason we can now talk from my iPhone to Android is because of interoperability. uh, very very important point. Um, uh, Mr. Khan, uh, I understand the, uh, immense potential that AI prevents presents in classroom for teachers and students. However, I know there are concerns when teachers utilize tools, uh, that they are not experts in, and when students employ AI outside the educational uses.
▶ 1:35:02Uh, I think you kind of touched on this a little bit earlier, but, uh, how can schools better protect and prepare educators and students, uh, to use AI safely and, uh, constructive purposes on scale? We believe that it starts with, uh, training educators, investing in educator literacy around generative AI, both, uh, you know, how the technology works, uh, the risks and benefits of the technology.
▶ 1:35:26Um, uh, an informed educator, one who is equipped to answer questions for students who are in, you know, are experiencing this technology whether they've chosen to or not, uh, through, uh, platforms, not only like the generative AI platforms that we're talking about today, but also AI as it's embedded in their daily lives on platforms that they might not associate with this AI revolution, things like TikTok, Instagram, you know, how are algorithms, uh, you know, incentivized to keep you on the platform?
▶ 1:35:56Uh, you know, how are you experiencing generative AI in your daily Um, I'm a big believer that a rich life is not staring at your cell phone 24/7. A rich life is engaging in the world. And sometimes when we can educate educators and they can inform students on how, you know, sometimes involuntarily you're finding yourself glued to a screen, uh, and it's because of AI algorithms that are powering this.
▶ 1:36:18It empowers the student to make better choices about how they engage with Um, and, uh, we we think that that it starts with educators and and bringing that to the forefront of an education. Technology is so embedded in our lives today, and, uh, often, uh, as mentioned, it's involuntary. So, we need to give students and teachers the information that creates agency for their own lives. Okay, thank you. Uh, Mr.
▶ 1:36:43Frazier, uh, your in your written testimony recommends that Congress thoroughly examine AI uses across the the economy to create evidence-based policies. What are some of the recommendations that you would offer to strengthen our, uh, data gathering Thank you very much for that question, Congressman. I think first and foremost, we need to make sure we're gathering data on a more frequent basis.
▶ 1:37:05We're seeing that AI tools of greater degrees of sophistication are being deployed at a faster and faster pace, and so making sure that our data collection keeps up or at least attempts to keep up with that space of AI development makes, uh, tremendous amount of sense. I'd also say making sure we're collecting data from across the entire private sector. Right now, there's a lot of focus on how major Fortune 500 companies are using AI, but we've long known that small businesses are the main driver of economic opportunity and job creation in our country.
▶ 1:37:35And so, better understanding the fact that, for example, right now just 9% of small firms are using AI. We're missing out on tremendous potential to help these entrepreneurs and innovators scale up their opportunities and hire more Americans. And then finally, I would make sure we're doing it in a way that's standardized and anonymized. Companies should not be punished for championing innovation and by leaning into AI. And so, we need to make sure this data is stored and maintained and shared in a way that champions innovation and doesn't punish those leading it. Thank you, and I agree.
▶ 1:38:05The, uh, small business owners, that's where we drive our culture. So, we need to make sure they're part of this process. So, thank you so much, and I yield back. I thank the gentleman, and I recognize the gentlelady from North Carolina, Ms. Adams. Thank you, Mr. Chairman, and thank you to, uh, witnesses for being here. Um, artificial intelligence is already reshaping our classrooms and our workplaces, whether we're ready for it or not. Students are using it, employers are are using it, and the federal government is still trying to catch up.
▶ 1:38:35Uh, in North Carolina, universities are racing to launch AI majors and research centers, not just to to meet the workforce demand, uh, but out of fear of being left behind. That urgency raises an important question. Are we building sustainable pathways for students or simply chasing the next academic gold rush?
▶ 1:38:56Uh, today I I I want to recognize, uh, this, uh, Virginia McCauley Jacobs, who, uh, is the CEO of the National School Boards Association in the audience. I thank her for being here. She's a leading She's leading the charge to ensure that school board members are also prepared to create AI policy in in K-12 schools. And thank you so much for being here. Building an AI-ready America isn't just about speed. It's about trust.
▶ 1:39:26It's about fairness. And who gets left behind when decisions are automated without guardrails. And that's where this conversation matters. So let me just turn to our witnesses, Ms. Ms. Reeve Givens. Um a recent survey found that 62% of higher education leaders estimate that fewer than half of their faculty are using AI tools while students are adopting them rapidly, often without guidance.
▶ 1:39:55And from your your perspective, what risk does this uneven adoption create for students and educators? And And how should institutions and the federal government think about supporting responsible AI use without forcing adoption or or lowering academic standards? When we think about how to make sure that AI brings everybody along across our society and economy, I think there are two different ways to think about this. One is about equitable access.
▶ 1:40:25Who can use these tools? Who are getting the skills and the benefits? And two is making sure that these schools aren't unintentionally creating greater divides either through discrimination or bias or harmful uses of of this technology. And we need to make sure we're focusing on both. So the first about equitable access, our research shows that even as we studied K-12 schools, 85% of teachers and 86% of students are using AI.
▶ 1:40:50There are big discrepancies when you start going in on a geographic basis and on an income level as well. And so how do we think about the digital divide and I guess addressing those gaps? On the difference that you described in colleges of students getting more familiar with this technology than their professors may well be, all of this comes again to the notion of how do you prepare educators to be teaching the workers and the citizens of tomorrow? And realizing that they do need to become familiar with this technology.
▶ 1:41:17And again, the point I made earlier, this is not just about the tips and tricks to make AI help you do your job better, although that is very important. It is also taking a more sophisticated understanding of how AI is transforming learning and mitigating potential risks.
▶ 1:41:30Right. Thank you. The The Center for Democracy and Technology has been been tracking the use of of automated decision tools, systems that make high-stakes decisions without hiring, promotions, or access to credit, often without human review. So what concerns you most about the use of these tools in the workplace? And do you believe that the federal government has a role in setting baseline protections so that that automation does not quietly reproduce discrimination or shut out workers in terms of opportunity?
▶ 1:42:02The use of AI in hiring and other decisions about employment is rapidly proliferating. And of course, there are efficiency reasons why employers are motivated to use these tools. But if the tools aren't rigorously examined and tested, there are real risks of them perpetuating discrimination because they are learning just on the historical data of who's been employed or also perpetuating snake oil. In my testimony, I talked about an AI video interview. This is a real example where the person spoke in German instead of English and was still deemed a good match for the job.
▶ 1:42:32This tells you that employers don't necessarily know what they're buying. Workers definitely don't know what they're being tested about. And so transparency, rigorous testing, and then avenues for redress are hugely important. Thank you. So quickly, how should policy makers think about preparing students for an AI-driven economy in a way that emphasizes AI literacy, critical thinking, and ethics rather than narrow technical skills that may quickly become obsolete?
▶ 1:42:58Um I'm not an educator, but my impression is parents, teachers, and students care about this a lot. And one of the things we need to focus on is realizing AI is a tool. You are still the person that needs to be engaging in critical thought. And so understanding that comprehensive approach of when AI can help, but also its inherent limitations and the need for safeguards to make sure that it's being used responsibly and well, that is the complete package that we need to make sure teachers are being resourced to provide to their students as well as of course using these tools responsibly themselves
▶ 1:43:29when they're using it in an education setting. Great. Thank you very much. I'm out of time, Mr. Chairman. I yield back. I thank the gentlelady and I recognize the subcommittee chairman of early childhood, elementary, and secondary education, the gentleman from California, Mr. Kiley. Uh thank you, Mr. Chair. I wanted Ms.
▶ 1:43:46Nayak to ask you a little bit about the certification program that OpenAI I has, which I think is is really interesting in equipping people with um uh you know, the skills to effectively use AI tools in the workplace. It seems that one of the challenges you face is that you know, the capabilities are continually increasing. And you know, you guys are always putting out new models and new features.
▶ 1:44:12And so you know, what a certification certifies today may you know, not necessarily reflect um what use can be made of your tools you know, a week from now, a month from now, a year from now. So how do you manage the certification program in light of the ever-increasing capabilities of your Congressman, this is something that we talk about quite a bit internally and there's two modes that that we're going to use to address this.
▶ 1:44:42One is you're you're certainly correct that you know, we release new models, we release new tools and technologies pretty much on a weekly basis. We talk a lot internally about how we can make sure that our courses can actually adapt as we actually introduce these new tools. AI is actually helping us do that where we're working rapidly to leverage AI in building these course works just like we put it in the hands of teachers to be able to build these courses to be able to adapt as fast as possible to the new products that we're putting out.
▶ 1:45:10So that's number one is actually having the course grow with the technology. Number two is when we think about actually giving someone a certification, how do we make sure that that certification follows them as technology changes rather than you know, they get the certification once and then they're done and then three months from now they no longer have that skill.
▶ 1:45:30The way that we're thinking about it is unlike certifications today, which are course that you take and an post that you get and then you get that certification, our certifications will actually grow with the user as they use ChatGPT. Our North Star there, we're not there yet, but we're hoping to get there as fast as possible, is actually enabling people to get certified as they're leveraging the platform. So that certification will actually assess over time that they're actually building the new skills as the technology is changing. And so we're really hopeful about that.
▶ 1:45:59We think that we can use AI as a tool to actually grade and certify people along the way so that it is actually pretty nimble to the technology as well. Yeah, very interesting. And I know what a challenge that must be. You talked about the capability overhang when it comes to adoption in in industry.
▶ 1:46:17I think there's also something maybe more extreme overhang when it comes to the policy side of things where a lot of the debates we have are about you know, a state of the technology that we've already So on the education subcommittee, we've been looking at AI in education and I'm very optimistic about the opportunity I'll bring you into this as well, Mr. of AI in in education to close achievement gaps and to really transform learning.
▶ 1:46:44But you know, sort of by definition, we're only able to look at what current tools are being used to do. I'll start with you, Ms. Nayak, and then would love to hear your thoughts, Mr. Does OpenAI have a vision for sort of five years from now given what the capabilities, the new modalities we might have, the different interfaces rather than this the chatbot interface? What education might look like in an AI you know, with the progression of AI And then I'll go I'd love to hear your thoughts as well, Mr. Khan. Thank you, Congressman.
▶ 1:47:15Yes, so we think that AI will augment the systems and structures that we have in place and ideally make them a lot more personalized and a lot more focused on the individual learner. And I think that we're already certainly seeing this with efforts like Magic School. And so what we're doing is is in a couple of lanes. One is that we actually need to teach teachers how to use our tools so that they can actually design curriculum that leverages AI and help their students learn AI as well.
▶ 1:47:43If we do that, then we can have certifications courses or even students can bring their curriculum from their schools into AI and they can get personalized support. And we think with that, we can actually grow the ability of teachers to impact so many more students because it's like every student can get a personalized tutor. And I think in the past, you know, that was only accessible to the select few who had the means to be able to do that. With workers and with students, again, we think that AI will be able to support them in personalization.
▶ 1:48:14So we're already seeing this trend. 30% of conversations today are people actually going to ChatGPT to learn. And that's across workers, students. What we see is that we're actually allowing access to education at a level never seen before. And then if we can capture what people are learning as they have access, we think that we'll have kind of a microscope into what people actually know in a way that we never had before. I know we're running short on time.
▶ 1:48:40I think you know, the example I gave earlier around personalized writing feedback for every single student at scale can move outcomes for students. We've seen test scores improve when a student gets personalized writing feedback rather than you know, in the moment that they finish writing something rather than waiting two or three weeks for their teacher to get them feedback on the essays that they've given. So we have this personalized learning AI tutoring type aspect that is true today.
▶ 1:49:05I think the way schools look, you know, 5 years from now is a lot more project-based where students can be equipped with generative AI tools and be centered in the classroom as builders. They're building and manifesting their own creative work because they have these tools of the future alongside them. They can build their own apps. They can help their What my dad was an HVAC technician, and I would have helped him build his website for his business and maybe the CRM that helped him to keep track of his clients. So, lots of builder experiences we see in the Thank you.
▶ 1:49:35That's very exciting. We get it right. I yield back to the chair. I thank the gentleman. I recognize the gentleman from California, Mr. DeSaulnier. Thank you, Mr. Chairman, and thank you so much for this hearing. I'm filled with possibilities and anxiety about our testimony. I appreciate all of you. I also want to point out to the chair and thank you for two of your witnesses are graduates of the University of California, Berkeley, and neighbors. So, don't know if you were We We planned that.
▶ 1:50:03Yeah, okay. And Ms. Naik is a constituent who doesn't live far from me. Um so, in the context of the apprehension, uh Pew says that one in five American will be displaced by AI. Um I think RAND Well, I know RAND did a study that said one in eight adolescents asked for mental health advice through uh bots.
▶ 1:50:28And that 40 50% of students feel uh disconnected from their schoolroom environment and their teachers because of AI. So, that creates my apprehension. Um a proud representative of a Bay Area district, which happens to be the fifth wealthiest in the country. A lot of that driven by technology and people like Ms. Naik who um live where a lot of this is happening.
▶ 1:50:53But, I also grew up in Lowell, Massachusetts, listening to my grandparents tell generational stories about working in the mills and technology and both how that was beneficial, but it took a long time for America to catch up to some of the downfalls and enact minimum wage standards and child labor laws. So, I feel like we're in one of those moments. I have a lot of young people uh tell me about ghost jobs. Um so, Ms.
▶ 1:51:21Givens, first, are we remotely prepared, in your opinion, sorry to use that word, um just in terms of our social safety net, uh workforce development, and I can see it happening right now when I hear young people talk about ghost jobs that they just they're applying and the algorithms are just dismissing them, and we know from research that up to half of those jobs aren't real, that they're just people collecting data, and because social media is protected by section by being immune
▶ 1:51:52from section 230, amongst other reasons, this is exponentially getting worse. So, ghost jobs and how young people are so disaffected um from just trying to get jobs, particularly people who have taken on large student debt because they've been promised if they get a good education, there's great opportunity for Uh thank you for the question, and uh I think one of the things to focus on here is just the incredible rate of change.
▶ 1:52:19Right. It took over two decades for personal computers to be widely adopted. The adoption of generative AI by people has been significantly faster, more, you know, over a billion users in the first 4 months of rollout in ways that speak to the promise of the technology, but also the speed with which we need to Pulling on a couple of the threads in your question, this this role of, you know, AI in hiring. So, yes, ghost jobs, one part of the problem.
▶ 1:52:43Also, people, you know, you throw your resume into a pile and have no idea how you're being evaluated even when it is a real employer at the other end of the funnel, but how are you being evaluated? How are you being assessed? These video interviews where you're just being interviewed by an AI system without having any sense of of what the outcome is going to be. That is one significant piece. It's also a clear intervention point. The US has employment discrimination laws. We have robust laws here, but we need a lot more transparency for those laws to actually be enforced as well as a functioning Equal Employment Opportunity Commission.
▶ 1:53:13The other part that I will add is the the the world that Professor Fraser was describing about gigafication and what the future of the economy is going to look like. There are real tradeoffs when you think about career pathways, what professional development looks like, the community and identity that we get from being part of a workplace, and of course the questions around worker power. So, what does it mean to have a fair wage and to be able to negotiate when, as is already happening in some places, you're bidding for a job, you're bidding for a shift.
▶ 1:53:43The person that agrees to either take the most immediate uncomfortable time or to be paid the least is the person that gets that job. So, this is where our social safety net and our employment laws really have to accelerate rapidly to be able to keep up with that new world order and make sure that we're thinking not only about productivity gains for the economy, but what does a resilient, self-sufficient, independent workforce look like? What is it for workers to have dignity, and what are the safeguards that we need to help empower them in that way? Ms.
▶ 1:54:11Naik, as you're a constituent, I want to ask you a question. Um I think of Emily Bender's work from the University of Washington. Her book is titled AI Gone, and she gives the history about how private equity gets most of the benefit historically in our country. So, I've worked with the tech companies. I've worked with Facebook, but the promise in not inhibiting innovation, that challenge I've always felt, uh but self-regulation hasn't worked either. So, could you address that very briefly? Thank you, Congressman.
▶ 1:54:42Yes, I I We do agree that there could be a capability overhang where you do have this distance between the haves and the have-nots that can leverage AI versus those that cannot, and we think we're actually at a very critical moment to address that overhang, and that certifications and the ability to upskill people to be able to leverage AI, whether they be a small business owner, whether they be a student, um whether they be, you know, someone working in bed tech, um is critical, and that's what we're invested in.
▶ 1:55:11So, I think the the summary is that yes, you're 100% correct. We need to make sure that this technology is accessible to everyone, and we think one of the ways that we can do that is by providing them the tools to learn about it, to use it in their everyday lives, and then hopefully to profit from it, um not just, you know, the large corporations. So, now that I know you're a neighbor, we'll finish this conversation. You probably will regret meeting me, but Thank you. The gentleman's time has expired. I now recognize the gentleman from Missouri, Mr. Ander. Thank you, Mr.
▶ 1:55:40Chairman, and uh thanks to all the witnesses for appearing before the committee today. Um I'm very interested in what AI might have to offer to improve education. Of course, AI I think has an an enormous potential in various areas of our economy to improve productivity. For example, as a physician, um one of you alluded to radiology.
▶ 1:56:02Pathology is another area where AI can make the radiologists and pathologists more productive and, you know, and and let's face it, not miss um some things that if it were just the naked eye, um as well as decreasing administrative burden. It seems like more documentation is required by the by the year to uh to make the electronic health records happy.
▶ 1:56:26Um but, I am concerned though about potentially unintended consequences of AI, specifically in education, because we are dealing here with young people. Um you know, a teacher recently told the New York Times, quote, "The amount of time my students are inclined to work on something that challenges them has become much shorter over the past few years. There was a time when a typical student would wrestle with a concept for days before getting it.
▶ 1:56:50Now, if that student doesn't understand something within minutes, they're more likely to give up and look for an alternative, i.e., asking a chatbot." You know, the uh ranking member Scott cited an article from NPR uh today. Um it was citing um a study by Brookings.
▶ 1:57:10Brookings um uh did focus groups around the country around the world, 50 different countries of parents, students, educators, and tech experts, um as well as a literature review of hundreds of research articles. And um it it found some positives and negatives. One of the positives was that AI can help children learn to read and write, and that's of course great.
▶ 1:57:34Uh at the same time, um the study uh concluded in part uh AI poses a grave threat to children's cognitive development. Um the concern was a quote doom loop of AI dependence where students increasingly offload their own thinking uh onto technology, leading to a kind the kind of cognitive decline or atrophy more commonly associated with the aging brain.
▶ 1:58:01And they commented other technologies can do that, too, um uh for instance, word processors, um but it it's on steroids with AI. You know, in my concerns, I don't mean to pick on pick on my AI experts here because I have similar concerns about smartphones, about about social media.
▶ 1:58:18You know, Steve Jobs in 2010 was asked by the New York Times, "How do your How do your children like the iPad?" He said, "My children don't have an iPad." And we know many technology um uh executives, Bill Gates, Neal Mohan, uh others have have have limited their kids' use of of technology. But, I I'd ask I'd ask my witnesses, I'll start with you, Ms.
▶ 1:58:39Naik, uh how do we harness the potential benefits of AI in education and yet protect our kids from the risk of of overdependence, and of um I even have concerns just if the if the computer's in front of the child every day, you know, 20 It seems almost 24/7. How do you How do you protect kids from all the uh horrific stuff we know is on the internet. Thank you, Congressman.
▶ 1:59:07First I want to start that OpenAI agrees with you that we do really need to invest in protecting our youth and making sure that they use our technology for their benefit versus their harm. And we're putting a lot of work that's not my area of expertise, but we are putting a lot of work as a company into making sure that there are those guardrails. I do want to address kind of your statistics on AI usage and and the harm it could cause. We're also seeing that it could cause immense benefit.
▶ 1:59:33We're seeing that 1/4 of 18 to 24-year-olds use AI every day to help with their learning. That's about just general questions, tutoring, schoolwork. We're also seeing that for instance a study from Harvard University found that a physics class that actually had a tailored curriculum that was focused on using ChatGPT found that students actually doubled their engagement and actually improved their problem-solving skills.
▶ 1:59:58And so I think you're right that if teachers inadvertently kind of have students using AI there could be ways where they cheat or they use it in in ways that weren't kind of meant for the course. No, thank you. But if teachers can actually augment their curriculum to integrate it, I do think that there's promising ways that they could actually increase engagement, increase critical thinking and problem-solving.
▶ 2:00:19And I do want to also point out I I continue to be concerned about political and ideological bias in AI. Of course President Trump on July 23rd issued an executive order preventing woke AI in the federal government, but as of this week, in fact as of yesterday, I did an AI search and I was very I was pretty darn clear about what I was looking for and it returned results biased against what I was looking for.
▶ 2:00:48So unfortunately I'm I'd love to hear your response, but I'm out of time. Thank you so much and I yield back. I thank the gentleman. I recognize now the educator from Connecticut, Mrs. Thank you and thank you to our witnesses for being here today. Advanced technologies like AI are rapidly being integrated into the classroom and can enhance teaching and learning.
▶ 2:01:13It can also be difficult for teachers and administrators to navigate evolving AI technologies and to use them safely and effectively. Ms. Reed-Gibbons, in your testimony you state that "Fewer than half of teachers and students report they have received some training or information on AI use in class, well below the reported use of this technology." Mr.
▶ 2:01:36Khan, in your testimony you state need support to understand where AI is helpful, where it can fail, and how to supervise its use effectively." I agree with both of you and I was listening just now to you, Ms. Nayak, when you talked about teachers getting training to get certified in AI. I have to frame my questions with the caution that while I understand this is important I mean, I'm a certified classroom teacher.
▶ 2:02:07We had a a component in special education. Now a layer on AI and we're having conversations where we don't even want to pay teachers what they deserve to begin with but continue to add on all of these layers. And I did chuckle a little, Mr. Khan, when you said students have to wait two to three weeks for feedback on an essay. For the record, no student in my class ever had to wait two to weeks for feedback on an essay.
▶ 2:02:32And that time where I gave feedback, not just what my comments were, but why I made them in a one-on-one setting was valuable and promoted student interaction. I would set a day aside and either call students up to my desk one at a time or go around the classroom and talk them through the edits, not just make edits on paper. But I understand that this is technology that we have to embrace and kind of learn and grow with it.
▶ 2:02:59I do have serious concerns though about the expectation of classroom teachers to integrate these things. So my question Reed-Gibbons or Mr. Khan, whoever, what are the most beneficial professional development and training tools to support teachers and administrators in effectively integrating AI technology in schools?
▶ 2:03:22And where do we need to allocate more attention or resources to support the educators who will now be tasked with not only the technology, but the integrity, the ethics surrounding it and all of those components that come with it.
▶ 2:03:37Because when I went all of my degrees are in education curriculum and instruction, educational leadership but this is literally a whole area all by itself that [snorts] classroom teachers are going to be expected to be responsible I thank you for the question, Respectfully, all of my degrees are also in education. I was also a high school English teacher.
▶ 2:04:05So you get it. And I do I do get it in public schools and recognize that you know teachers already have a immense amount of burdens and training that they are already needing be equipped with in their daily work as educators. I think to to answer your question, the things that they can benefit the most from are you know the way that we work with schools and districts that we partner with is first we give them kind of a baseline education on how this technology works.
▶ 2:04:34Here are some suggested use cases on how this could be used powerfully. Here are the risks associated with those things. And we also empower the local districts' leadership the champions, maybe the technology leaders in the district, maybe a department chair in English. And they can talk about how to your point, you know, it is still really important that the student is hearing from the teacher about, you know, the how the strengths of their essay, the areas that they have to grow. And how can AI be a part of that story rather than the full
▶ 2:05:03how often would that training need to be updated because this technology is evolving so quickly? Yeah, I don't think Literally, you'd have to have professional development every week. Yeah, I don't think it's a singular training. I think it's just like you in our work as educators, we had department meetings every every you know, for for me in the in the school that I worked in, we had a department meeting every month. And here's what's new, here's some things that you can use in your classroom.
▶ 2:05:27But I think forcing this on teachers in terms of how you utilize it in your classroom has never worked top-down. In terms of a baseline education around it, helping you understand what you can and can't and shouldn't do, and then equipping you with the agency to make your own choices as a teacher is I think something that I've always advocated for as a teacher. We trust teachers. We empower them with knowledge and then we trust them to make good decisions. That's what I would encourage.
▶ 2:05:55Well, I guess just like 5 minutes is not enough time to question witnesses on this I don't think we are adequately positioned to require teachers to be responsible for this in the classroom. I'm sorry, Ms. Reed-Gibbons, I've run out of time. I I yield back. I thank the gentlelady. I recognize now the gentleman from North Carolina, Mr. Harris. Thank you, Mr.
▶ 2:06:16Chairman, and thank you to all of our panelists today and the witnesses for your expertise and what you've been able to share and some very important insights I think and a lot of food for thought that we've we've heard Mr. Fraser, I want to just start with you. I hear predictions, it's been mentioned even here today that AI could replace millions of jobs in the next decade.
▶ 2:06:39Some lawmakers and a number of pundits say that we should impose a tax on employers who adopt AI to replace workers. Now I know from my research you have publicly opposed this so-called robot tax and I want to just take a moment and give you an opportunity to explain why you think a robot tax is a bad way to protect American workers. Thank you for that opportunity, Congressman.
▶ 2:07:05I think that a robot tax is very well-intentioned like a number of economic policies, making sure that Americans, as we've heard throughout this hearing, thrive in an era of new technology. But the American way has never been to be scared of technology no nor to prevent investment in the technologies of the future. If we want to not only create a thriving economy, but also to realize that we're competing in an international space where we're not seeing China, for example, impose a tax on its robots. What are they doing?
▶ 2:07:34They're investing in creating the next generation of robots that are going to lead to new innovations, new economies, and new sectors. And so on the domestic front, we need to similarly recognize that when productivity increases, yes, we may see differences in whether that allocation is going more towards capital, more towards labor, but a net increase in productivity over time increases the whole economic pie. And that's what we should be focused on.
▶ 2:08:00There's no clause in the Constitution that says everyone has a right to the specific job they were born in. Just ask the horse pooper scoopers, right? They don't still walk around the streets of New York cleaning up. So what we need to focus on is creating mechanisms and pathways for opportunity. Do you have any specific ways that you might suggest that Congress, steps we could take to protect workers instead of this tax?
▶ 2:08:24So first and first and foremost, I think that we need to empower American workers by moving towards a portable benefit system. Right now there's a real sense of what economists will refer to as job lock. Folks feel stuck in their job because that's predominantly where they get their benefits from. If we want to make sure that Americans can thrive in this more fluid and dynamic economy, then it's important that they're able to take their savings, their time off from job to job. And in that way we can facilitate that economy. Got you. Thank you very much. Mr.
▶ 2:08:55Khan Um, know you've had direct engagement with the White House AI Education Task Force, which has made a notable effort to thoughtfully engage educators, parents, and federal uh leaders. In fact, uh in July of 2025, as already mentioned earlier, the administration released America's AI Action Plan.
▶ 2:09:15I want to just give you a few moments from your perspective, what stood out about how this administration is approaching AI literacy and safety, especially in preparing young people for the future? Yeah, in in the committee, I heard uh the folks around the table really focused on protecting children.
▶ 2:09:34Uh there were there was a lot of conversation around the idea um of how we've seen technology affect children in our in in recent memory with uh things like the way that social media has affected children in negative ways. Um the way that we've now seen, you know, full circle schools banning cell phones in schools. There was a remark made about uh how we we don't think that kids should be staring at their laptops every single day. But we think that technology is inevitable.
▶ 2:10:04Uh it it will be in our children's lives. So, um you know, there there's a lot of conversation of how can we make sure that both educators and families are equipped with the knowledge to make decisions about how they use technology. Uh you know, freedom is only freedom when you have information, when you have agency, when you understand something. Um and and that's what I heard the focus being on in that meeting, uh which I was heartened by. Excellent.
▶ 2:10:33Well, thank you very much and uh to all of you uh for your time that you've shared. And Mr. Chairman, I'll yield back. I thank the gentleman. And I recognize the gentlelady from Arizona, Ms. Grijalva. Thank you, Mr. Chairman. Um the rapid advancement of AI necessitates an acknowledgement of the profound ripple effects this emerging technology creates across society.
▶ 2:10:57While these issues may transcend the immediate scope of this committee, we must address the exploitation of intellectual property, privacy violations stemming from mass data collection, and the spread of misinformation. Furthermore, the integration of AI introduces significant security risks, a lack of algorithmic transparency, and severe environmental consequences due to excessive consumption of energy and As we navigate these changes, it's essential to remain vigilant regarding the risks
▶ 2:11:27and inequities that AI can introduce into our foundational systems. In Tucson, Arizona, um we had a proposal for an AI site, and overwhelmingly the community rejected it and will not be being built within um Tucson, Arizona. But it's moving on to other areas. And so I think it's important for us to think about all of the the um the things I just mentioned.
▶ 2:11:52Pima Community College, also in Tucson, Arizona, serves as a critical workforce pipeline for first-generation students and working adults. Through partnerships with regional employers, the college provides the job-aligned training required for a landscape where AI is reshaping everyday work. To ensure a resilient future, we must support these institutions in delivering the skills necessary for a modern economy while remaining focused on the risks inherent to um in the hiring process and in the classroom.
▶ 2:12:21So, this would be a question for Ms. Reeve Givens. Um what specific federal investments would most effectively enable community college colleges like Pima to quickly expand AI-related certificates across fields such as healthcare, manufacturing, and IT? And how can institutions ensure they are balancing the potential benefits of these tools with the risks of using AI for sensitive educational tasks such as tracking student progress or developing individualized education programs?
▶ 2:12:51Um thank you for the question. Mr. Harris alluded to some of the energy that is coming out of the current administration, including a focus on ramping up AI adoption in education. That creates an important opportunity provided we are thinking about distributing that opportunity fully and equally across the country and reaching communities that really need it most. It's also an important moment to think not only about AI adoption, but responsible AI adoption.
▶ 2:13:15In 2025, the Department of Education put out respon- uh principles for responsible AI use that have a lot of good content in there. We have advocated for the administration to follow those own guidelines put out by its own Department of Education uh in the rollout of those plans and opportunities. Um the other piece that I would add here, you were very right in your opening to touch on some of the other issues have not raised by the committee yet today. Cybersecurity threats that schools are facing, the privacy concerns. Just a pause for a moment on the cybersecurity threats.
▶ 2:13:43In our recent research, 23% of teachers reported that their school experienced a large-scale data breach in the last school year alone. Cybersecurity analysts report 4,000 schools experiencing data breaches in a particular week. And as AI use increases, we're sharing even more data about students, even more personal information that becomes vulnerable to that harm.
▶ 2:14:05Yet right now the administration has pulled back its funding for CISA, the government agency that supports this work, and ended the multi-state information sharing program that helps schools uh school districts respond to these threats. So, those are the types of places that might not be top of mind, but really need to be when we think about school readiness, school preparedness, and making sure that educators, again, not only are learning about the potential of AI and thinking about its use in the classroom, but actually are equipped to navigate this transformation and get the full benefit that they deserve.
▶ 2:14:35Thank you. Did anyone want to add anything to to that question? Um I don't know I I what safeguards should Congress include to ensure AI workforce investments are broadly accessible to community colleges outside of major technology hubs like Pima in Tucson.
▶ 2:14:53And given the active litigation against Workday involving allegations that algorithm-based screening tools systematically discriminated against applicants, what can be done to the algorithms underlying these tools to ensure that they do not have a disparate impact on our graduates because of their age, disability, race, gender, or other protected characteristics?
▶ 2:15:15Sure, I'm happy to talk briefly about some of the the lessons from uh a litigation that Workday is facing right now about discrimination through hiring tools that it was used without specifically commenting on the merits of that case, which um continue to be ongoing. It's a really important wake-up call not only to employers about how these tools can perpetuate discrimination, and also the role of the vendors that are developing these tools as well. Um this includes questions like how easily and rapidly people are being dismissed based on protected characteristics.
▶ 2:15:46Um one of the things that I think is important for Congress to know is yes, we do have existing employment laws that can be brought to bear in these cases, but it is very hard to actually discover these instances and to prove these cases because of the lack of transparency and visibility. It used to be there was a bad apple in HR, and you could maybe trace back how an interview went badly, or you know, what might have happened just from studying the results of one company. But here, when you're just uh as the congressman was saying earlier, throwing your resume into a void with no idea what's happening next.
▶ 2:16:16In the Workday case, a person submitted hundreds of resumes and had them all rejected almost instantly. And that was how he began to connect the dots. But what we really need is transparency around what these tools are assessing, how they're being used, and then an empowered Equal Employment Opportunity Commission that can be a cop on the beat to protect workers and make sure that employers are making responsible choices when they're deploying these tools.
▶ 2:16:40I thank the gentlelady.
▶ 2:16:41Thank you. I yield back.
▶ 2:16:41Time is expired. I represent I recognize the gentleman from Florida, Mr. Fine. Well, thank you, Mr. Chairman. And you know, usually when I prepare for my committee meetings, I do research, I understand what's going on. But today I had an expert that I could go and talk to about the use of AI in education, and that is my son, Jacob, who's a senior in high school. And who actually uses these tools and uses your product um quite a bit. Not to cheat, at least that's what I think. But but I will tell you as a parent, I have to ask him like, "Wow, that was really well written.
▶ 2:17:12Like, did you write it yourself?" And and he often gets upset. But he has shown me himself tools that none of us had, and frankly even students 5 years ago never had. And so, the first question is a question he asked me to ask. So, I'm interested in your perspective on this. He says AI can be very helpful to him in a whole variety of ways that you've focused on.
▶ 2:17:32However, because of negative effects of AI and other technologies, particularly social media, schools are now banning the use of technology in class. Like, he cannot use his phone during the school How can you leverage your product, build your products and technologies to sort of allow it to to push back and say, we're going to come up with ways where you can use our technol- the good stuff in school, but keep
▶ 2:18:02the bad stuff out." Are there ideas that are being bandied about in order to be able to do that? Whoever wants to answer. Good, Mr. Khan look like looks like you're raring to go. Yeah. Uh so, uh we actually built our product in this way that doesn't mean that the child always has to have their laptop out in class or always has to have a device. Rather, there are specific use cases that the child can use generative AI in the classroom under the supervision of their teacher in a way that is uh prescribed by the teacher for a specific purpose and or use case.
▶ 2:18:33There are tools in the platform. Uh an example would be a teacher has a uh research paper that students are doing in their senior year. Uh and Magic School has a great uh AI research assistant tool that students can use as a uh a support for them as they're working through their research papers. A teacher launches that to their students, the student joins that uh activity, and they can now use this AI tool as a part of their assignment.
▶ 2:18:59They're still responsible for the final product, but they can use this tool that's been built with guardrails and formed by what the actual task is that they're asked to be doing in class, so that it doesn't write the essay for them, but it does assist them in the way that the teacher intended them to be used in kind of a specific time period with which the teacher has prescribed for them to use it, and then it can be closed up when it's not supposed to be
▶ 2:19:24Well, my advice to all of you is this. Really, the issue is smartphones more than laptops, where smartphones have simply been banned. You can't have them. And if I was all of you, I used to be a software guy, didn't have AI, I'd be trying to work with some of these providers to come up with modes where soft where phones can be disabled
▶ 2:19:44Mhm. to do the bad stuff, but they can be enabled to do the good stuff. Because right now we're having to throw out the good stuff because, frankly, the bad stuff is pretty bad. We don't want kids on Snapchat and TikTok all day, and that's the problem. They can't access the good things because of that. But my my my my other question is also actually is for you, Mr. Khan. So, you talked about how teachers can use AI to help grade and assess students.
▶ 2:20:08Um you know, you can understand how this would help teachers save time, but it's also easy to imagine how this can go horribly wrong having a computer assess a grade. How do you make sure that the AI tools are are fair and accurate when they're being used for things like grading? Yeah, first and foremost, we do not grade with the platform. So, we don't assign a specific letter grade and or a score for a student on an assignment.
▶ 2:20:32We we keep it to things like formative feedback for the child, um but we do inform the teacher with sometimes a suggested grade. So, the platform has an idea that it could suggest, if the teacher elects to get a suggested grade for an assignment. Uh the way we control for that is we require every educator to review that suggested grade. We we have kind of responsibility flags that are held up to the teacher saying, you know, you are the teacher of record in the classroom.
▶ 2:21:01If you've ever been a teacher in a public school, you understand this term teacher of record. It means that you are responsible for that child's grade in your class, and it's like your name signed to it. So, Magic School never supplant the teacher of record. They are the ones who are making the decision about the final grade. We think of that as a high-stakes decision. So, it's about educating teachers on how to use this and making sure that they're using it in a responsible way and not as a way to replace their own thinking or assessment of a child's work. Well, thank you.
▶ 2:21:29Well, look, I think the challenge for us is this. I often talk about certain drugs. You know, severe painkillers allowed us to do surgeries that could never be done before and has had extraordinary effect on people's lives, but misused, they have horrible effects and they're killing Americans. I think most new cutting edge cutting edge technologies fall into this situation where they can be used for real good or they can be used for real bad.
▶ 2:21:53I would say the challenge for all of you and for all of us is to figure out how we harness the good without unleashing the bad, and I know that's something we'll be continuing to work on. And Mr. Chairman, I yield back.
▶ 2:22:02I thank the gentleman. I recognize now the gentlelady from Illinois, Ms. Thank you, Mr. Chairman.
▶ 2:22:11With how fast AI is developing, it's essential that Congress develop policies that promote educational workforce and workforce development while upholding user safety. It's also a matter of national security that the US win the AI race. I appreciate the work that the Trump administration is doing using AI to ensure students are prepared to enter the workforce while preventing it from being used for woke indoctrination.
▶ 2:22:39AI has become especially effective in the classroom as it can personalize learning experiences while also helping students with special needs and learning AI is equally beneficial to teachers as it helps create lesson plans, saving teachers hours of work per week.
▶ 2:22:58Over the course of a school year, this adds up to nearly 6 weeks of saved work AI is also beneficial to businesses and their employees as it can teach valuable skills at a fraction of the cost. However, as AI continues to develop and progress, we must address how it may disrupt the job market as well as teacher-student relationship in the Um and Mr.
▶ 2:23:23Khan, I'm especially concerned about AI's effects on children um concerning chatbots imitating human relationships. How can we address these issues while still using AI effectively in the classroom? Uh thank you for the question, Congresswoman.
▶ 2:23:41Uh we we we share in your about the inappropriate relationship that a student can have with a chatbot in the Uh the way that we guard against that is we make sure that any interaction with generative AI through our product and the way we believe all products should be used or AI products when used in schools should be under the supervision of a teacher.
▶ 2:24:06Teachers have monitoring dashboards where they are flagged for inappropriate content or maybe inappropriate interaction that the child might be having in the experience. And then, of course, the the experiences themselves are academically focused. They're not intended to to create a relationship between a chatbot and a child. More importantly than that, the certainly there are there are technology safeguards we put into the products, but we equip teachers with understanding the risks that you just described.
▶ 2:24:35We help them before they even use this technology with their students in the classroom. They're introducing the technology with letting the child know about what this is, what this is not. This is not a companion. This is not something that you should be building a relationship with, and human connection is still the most important part of the educational process. Uh so, the center of our work, you know, we have a slogan, it's teachers are magic, but that's very much the ethos of the company, too.
▶ 2:25:03Is that the greatest learning happens with the relationship of a teacher and a child and a parent and a child, and we we keep that the center of our work and only use AI to augment their ability to serve the needs of the children in the classroom and avoid the negative effects. I'm glad that we agree on the danger that children could access that.
▶ 2:25:26I know a school in Illinois recently, the kindergartners were working on a Chromebook and pornography just popped up on it, and so, you know, we just need layer layer layer of protections to make sure our kids' minds are protected. Absolutely. Um okay, and then also for Ms. Nayak, how can Congress develop policies that will use AI to promote workforce development without sacrificing jobs?
▶ 2:25:55Thank you, Congressman or Congresswoman, excuse me. We think that Congress should think about how all of the programs that already exist, unemployment programs, programs to help upskill and educate people, they should think about how those can also add AI skills into those trainings.
▶ 2:26:13So, what we're doing is we're providing a foundation of certifications, we're partnering with governments, we're partnering with employers to think about how those certifications can enter into their workflows, to think about how those certifications can lead to jobs, and we'd love to work with Congress to think about how our certifications could also be used in workforce development programs at the state level as well as Would any of those certifications be available to students while they're still in high school so that Okay. Yes, that's our intention.
▶ 2:26:43Students, teachers, as well as the workforce. Okay, great. My final question is to Mr. Frazier. Oops. Chairman, you want me to finish? What Can you talk more about needed safeguards for AI policy in light of student privacy?
▶ 2:27:02Yeah, first and foremost, very quickly, I would say we also have to be attentive to the fact that today's AI is the worst AI we'll ever use, and if we want more AI tools to be safe and reliable in a student context, we also have to talk about updating and gathering student information. We can't have one without the other. So, better data leads to better AI. Thank you, and I yield back.
▶ 2:27:23I thank the gentlelady, and I recognize the gentleman from Pennsylvania who's wondering how AI will help produce better whole milk. Yeah, I'm the chairman of the Ag Committee, Mr. I've got a lot of thoughts on that, great applications in agriculture, but let's talk education today, um which is the foundation of everything, actually. So, Mr.
▶ 2:27:46Khan, you had reflected already on how K-12 schools currently integrate AI-based assistive technology in the classrooms. I want to I want to focus on students who are learning with various disabilities. What do you see as the opportunities there? We, you know, we uh you know, the reauthorization of the Individuals with Disabilities Education Act is long overdue, and so actually, this could be good timing.
▶ 2:28:16Uh thank you for the question, As a former special education teacher, this is near and dear to my heart. Uh AI can be an incredible assistive technology in supporting students with learning disabilities. Within our platform and many AI tools, one of the incredible advancements is in speech-to-text and text-to-speech.
▶ 2:28:40So, often students with IEPs, individualized education programs, IEPs, Uh with disabilities will have a read aloud Uh AI can read aloud in a natural voice that continues to have the child engaged in their education with a compelling um individually in the classroom rather than having a teacher kneel down next to them um to have to do that.
▶ 2:29:03And perhaps, you know, that's great for the individual child the teacher's with, but you know, when I was a a special education teacher co-teaching, I couldn't get to every child in the classroom to give them that accommodation. There was a physical barrier for me able to be able to you know, implement the required uh that the children in the classroom that I was supporting had. Um and AI uh breaks down that barrier. It gives them the ability to have read aloud in the moments that they need it.
▶ 2:29:28Um in addition to that, when AI is informed with uh the needs of the child especially when they're individualized needs uh there are are state mandated or or federally mandated through an IEP, then the interaction that they have with the technology can be even more personalized to the needs that they have. Another example I might give is uh you know, when I was teaching ninth grade English, uh there were students who were many grade levels behind in the class and students who were well advanced of uh their classmates.
▶ 2:29:59Um AI can give them personalized feedback across their ability levels if informed by uh their zone of proximal development. So, you know, that that's the idea that you don't want to give a child um you know, work that's so challenging that they disengage. You want to give them to them in a way that's like just stretches them just enough to help them grow to the next Uh and AI can help get them into that zone of proximal development based on no matter where they started.
▶ 2:30:28Uh so, when we can use the data that we have about kids and their needs, uh then AI can be something that increases the capacity of the educator to serve the diverse needs of the students in front of them in a way that's never been possible before. Uh that gets me really inspired. Special education's probably the thing that that, you know, is again near and dear to my heart, but also one of the most incredible ways uh that we can support the real needs of schools and districts right now.
▶ 2:30:55And you know, special education is a designation, but every child learns differently, right? And there are students who are uh multiple language learners in schools who can be served with this technology in translating in and out of English and their native language. Uh there's so many incredible applications uh for the specialized needs of learners in classrooms. Very exciting. Very exciting. Great for the kids, uh great for the teachers, as well. Um Ms.
▶ 2:31:23Nayak, uh how is AI technology being integrated into workforce development to enhance skill acquisition and support career transitions? Thank you, Congressman. Um we're seeing a lot of enterprises actually very excited for AI skill development. And um when we actually speak with our enterprises who are adopting our tools, one of the biggest things that they ask us for is actually certifications and skills-based training that they can provide to their workforce so that they can leverage these AI tools.
▶ 2:31:51And so, the way that it's being integrated is um at OpenAI, we're partnering with enterprises to actually build out coursework and curriculum and certifications um that will teach their workforce the skills that they believe that they need in order to be successful in their jobs in the AI era. And over the course of the next several months, we're going to be releasing several courses that we've built in partnership with these um organizations. Very good. And um and how is AI being used at this point in career and technical education or CTE programs?
▶ 2:32:24CTE is a a emergent use case in a really, really powerful way. We found that a lot of CTE educators are coming from industry. They're folks like my father who's an HVAC technician, welders. They didn't get a traditional training and education. So, they're learning how to use AI to lesson plan and and build all of these really incredible resources where curriculum doesn't exist to bring this technology in a way that helps folks build these uh working class or or what I would I would call living wage jobs.
▶ 2:32:55Right? They're they're getting access to these living wage jobs and the teachers of those jobs can actually uh learn how to be a teacher a lot faster because they're able to utilize this technology to build things like lesson plans as non-traditionally educated educators.
▶ 2:33:07Very good. Well, thank you. My time's expired. I call them essential workers, I thank the gentleman. And now, since we've uh discussed AI with whole milk, I suppose we ought to discuss AI's impact on bourbon. Absolutely. I recognize the gentleman from Kentucky, Mr. Comer. Well, well, thank you, Mr. Chairman, and I thank our witnesses for being here today. I'm very excited about what AI could mean for my district in rural Kentucky.
▶ 2:33:33Uh in my congressional district in Paducah, Kentucky, a former Department of Energy site that once supported nuclear weapons production is now being redeveloped for advanced energy projects and AI data center uses alongside uranium enrichment facilities.
▶ 2:33:48Just last week, I'm very happy to uh announce that one company in Paducah received a $900 million task order from the Department of Energy to enrich uranium and power America's nuclear At the same time, local workers are preparing to reskill and upskill to meet the needs of companies reindustrializing the Paducah Department of Energy site. For a lot of folks back home, this isn't theatrical.
▶ 2:34:16This is it's about whether they can stay their communities and move into good, well-paying jobs as technology changes. Paducah and the surrounding communities in West Kentucky have a skilled workforce built around energy manufacturing and national security, and they're now being asked to transition into this next generation of infrastructure. I see this as a real opportunity for Kentucky uh uh to be a part of the future of AI, not left behind. Uh Ms.
▶ 2:34:42Nayak, how do you see AI-driven industries creating new kinds of jobs in places like Paducah, Kentucky? Thank you, Congressman. And I think Paducah is a really exciting example for how AI could actually move within industries and kind of transition them over time. Um I think the example that you gave of um data centers and being able to leverage their skills to power the next generation of this technology is a very exciting one.
▶ 2:35:08Um what we're doing with the jobs and certifications effort is um we're actually leveraging our AI Foundations courses to build curriculum for specific industries like the ones in your district um to enable them to actually gain the AI skills, but also leverage the skills that they already have in this new um era. And um we're very excited to see that enabling workers to actually transition with the times and then um actually lead to much more economic growth in new industries or industries that are just transitioning with AI.
▶ 2:35:38Well, that's great. And and we talk about it a lot in this committee, and I think uh we're all in uh a majority agreement that it's a new type of of job training in America. And and I guess Ms.
▶ 2:35:53Nayak, uh what specific kinds of of training or or specific certifications would help average workers that want to reskill move more quickly into these new types of jobs? And how does OpenAI think about supporting those pathways? Congressman, we're investing in several different styles of courses um that will lead to an OpenAI certification.
▶ 2:36:21Our first one is um something that we think can be valuable to everyone from a teacher to a student to a worker, which is AI Foundations, which teaches them how to use tools like ChatGPT, teaches them how to build agentic workflows, and then also teaches them how to bring their community along and actually train the trainer and um teach others around them to use AI. Um that's our kind of first set of courses, and we're already starting to roll those out. Um we'll finish that full curriculum series this month. Um the second set of courses are focused on AI Builders.
▶ 2:36:51Um those are the ones who are technologists or even um like the examples that you gave who are operating data centers or operating energy facilities, where they're actually leveraging AI to build tools for others to use. Um we're going to start to build out an AI Builders curriculum and coursework that will empower those individuals to be able to use our tools. Um over time, we see that we would love to partner with every industry to think about how we can take those types of courses and then adapt them to their industries.
▶ 2:37:18Um so, examples of that is we've already released a AI um teacher course, which takes our AI Foundations course and actually augments it and implements it for teachers. Um we're hoping to do that in the healthcare space as well. So, we think it's very exciting. We think that this foundational curriculum can lead to a lot of benefit for many different industries. Well, well, thank you very much for your insight. I'm very eager, Mr. Chairman, to to see how AI can unleash workforce development opportunities for my constituents in Kentucky. So, thank you all again for being here.
▶ 2:37:49Uh Mr. Chairman, look forward to working with the committee to see that we can achieve this goal of of retraining a uh a new workforce to be able to quickly adapt into the the bright future that AI has. So, with that, I yield back. I thank the gentleman. I now recognize gentleman from Guam, Mr.
▶ 2:38:11Chairman Walberg and Ranking Member Scott, thanks for hosting today's important hearing, and thank you today to our panel for providing your So, the rapid advancement of AI deployment through the workforce and and it's provided untapped opportunities, of course.
▶ 2:38:30And while we have an unlimited potential, we should also be cautious about irresponsible use and the threats of our adversarial what they compose with the new technology and that's why I work consistently with my colleagues on both sides of the aisle to ensure that appropriate guardrails and strategic planning are in place to generate jobs, ensure AI dominance, and protect our nation and its people. Um, Ms.
▶ 2:38:59Naya, oftentimes certain areas can be left behind in innovative workforce development. To maintain America's AI leadership across the globe and in all corners of the country, what approaches can ensure that no community is left behind, particularly in the territories? Rural areas and remote locations where AI applications can solve critical infrastructure, logistics, and economic development problems.
▶ 2:39:28Thank you Congressman, and we we share in your concern that AI could leave people behind or it could actually help to accelerate um rural communities or communities that normally wouldn't be reached by technology, and we're hopeful that it will be the latter. Um, what you're talking about is something we call the capability overhang where we want to make sure that people can actually use AI to its fullest and we need to be able to scale them to be able to do so. And so, what we're doing here at OpenAI is um a couple of things.
▶ 2:39:54One is we're leveraging certifications and those certifications are available on ChatGPT and so really anyone could access them. The beauty of that is that with ChatGPT as well, you could actually personalize a course towards different communities.
▶ 2:40:09Um, so you mentioned territories, you could actually um take a course and you could translate it to different languages or um you know, different levels of um education or understanding and I think that that will actually make this more adaptable so that it can actually be personalized to those communities rather than um something that's only focused on kind of urban areas in um the US. Thank you for that.
▶ 2:40:31Uh, also another question for you as AI transforms uh like finance services, supply chain management, and health care, what mechanisms would you recommend for enabling colleges and universities to rapidly redesign programs and deploy industry recognized credentials that meet immediate workforce needs while students and families are demanding these changes? Thank you Congressman.
▶ 2:40:56We've been working quite a bit with universities, not only um kind of large universities, but also community colleges. Um, and we've been talking to them a lot about how we can leverage our certification programs within those universities. Um, one way is that we actually need to scale up um the professors and teachers in those universities, and so we're building coursework for those teachers to actually understand how do you build AI-based curriculum. Um, that's part one. Part two is we're building coursework for students.
▶ 2:41:24Um, right now we have a student AI lab where as we're building this course curriculum, we're actually having that lab give us feedback both positive and negative as to how it's actually helping them, where we're actually missing the mark, and that's been super useful um for us in developing this course to make sure that it actually applies to our students and universities.
▶ 2:41:42And then third, um we're also working with enterprises so that they can tell us where they're seeing the biggest gaps and then ideally we can take that back to the universities and community colleges and actually partner with them to close those gaps. Um, and we really see this as a virtuous loop and we're excited that AI can actually help create and connect that loop in ways we never saw before. That's terrific.
▶ 2:42:03Uh, and going on that line with the commercial enterprises, how can federal policy support small and medium-sized businesses, local governments, and public institutions like schools and public health systems in adopting AI tools and upscaling existing workforces, particularly where expertise and training is limited? Thank you Congressman.
▶ 2:42:25We're actually seeing that around 40% of small businesses are already reporting using AI today, and um the majority of small businesses actually see this technology as increasingly important to their industries. And so we're we're actually very excited about the impact that this could have on small businesses from um you know, businesses that couldn't afford to have specific services where they were kind of stringing together work themselves, um now maybe they could use AI to augment it so that they have to spend less time doing things that they don't love and spend more time building
▶ 2:42:55their business. Um, what we're doing here is um again with the certifications effort, we want to give small businesses the tools to be able to adopt AI really rapidly um and to be able to prevent that capability gap so that they can actually use it to increase um their businesses. Thank you so much. Uh, thank you to the panel. Thank you, Mr. Chairman. I thank the gentleman. I recognize the very patient uh gentleman from Virginia, the ranking member, Mr. Scott. Thank thank you, Mr. Chairman. Um, Ms.
▶ 2:43:25Givens, um we've talked about bias and you mentioned guidance on websites from federal agencies. Can you say a word about the status of that guidance? Um, absolutely and it comes up in both of the topics of today's hearing.
▶ 2:43:40So, in the hiring and employment space over the past 5 years, the Equal Employment Opportunity Commission, the Department of Justice, Department of Labor, and National Labor Relations Board did extensive work to put out guidance for employers and workers about how to use AI in way that complies with existing employment laws. Unfortunately, all of those pieces of guidance have been removed from their websites.
▶ 2:44:01I do provide links to them in my written testimony from people who preserved that record, but unfortunately that pullback is just a very troubling sign about the leadership those agencies are showing right now for workers and employers. In the education space, I'll just note that the Department of Education through its Office of Education Technology and the Office of Civil Rights also put out very important guidance on how to ensure the use of AI in education spaces complies with existing civil rights laws. It's important that that continues to be available to people as well. Thank you.
▶ 2:44:31And can you on that discrimination guide discrimination cases, can you talk about the importance of disparate impact analysis? Yeah, so one of the foundational tenets of our civil rights laws is acknowledging that one form of discrimination can best be captured through assessing the disparate impact that it has on protected populations or people based on their protected characteristics in our society.
▶ 2:44:56If that protection is removed, then anybody who is trying to say they've been treated unfairly under the law has to make a very difficult case about what was in the mind of the person that hired 99% white people into their workforce just to give one example. And that's incredibly difficult to do and it ends up essentially making our laws almost impossible to enforce.
▶ 2:45:17So disparate impact for a long time has been a tenant of how we actually address inequality in these key features of people's lives and making sure that people with protected characteristics that they are not being judged on that basis, but on their ability to contribute to a job in question. And if a business buys off-the-shelf software that turns out to be discriminatory, who is responsible for who is accountable for the discrimination? Under the law, the employer is.
▶ 2:45:43Right now it is difficult to bring those cases because of the lack of transparency and visibility into how those tools are affecting the workforce, which is why we need not only enforcement of existing law, but also additional protections provided by Congress to have there be more transparency and testing in when these tools are used and deployed.
▶ 2:46:01And in professional development, if um are there problems we need to be uh sensitive to in providing professional development with AI rather than the industry providing itself if there's embedded bias in the software? I think in any uh tasking and other training materials, it is important to think about what is being taught and the quality of what is being taught. Um, one hopes that that is the hallmark of effective learning and a an effective teaching materials.
▶ 2:46:29One of the areas we've been focused on in the ed tech space is how to make sure that ed tech vendors, just to give one example, are much more transparent about their grounding in learning science so that teachers, educators, school administrators can be making informed decisions when they turn to these tools and are asking not only how are you trained, what is the what is the science, but also how are you evaluating your impact over time. Some are starting to do this, but the paperwork around it, the disclosures, the standards as Mr.
▶ 2:46:56Com was saying, aren't there yet and it's an important area to focus on right now. Thank you. Mr. Frazier, you you pointed out the importance of collecting can you say what data we needed to be collecting on Uh, in 72 seconds, uh I think that there's a tremendous need to focus on adoption, what tools are being adopted, focusing on the intensity and novelty of that adoption.
▶ 2:47:21So not only do you have access to a tool, but how are you using it and in what ways are you using it? Then I also think we also uh need to pay attention to which tools are explicitly being used in which institutional settings. Getting an understanding of whether a company, for example, is using last year's iteration of a certain model or the latest technology can give us real insights as well.
▶ 2:47:43So, it's really uh a extensive effort that has to be maintained to get a true picture of how AI is diffusing across society, but it's one we must pursue if we want better
▶ 2:47:52Let me try to get another question in. Florida had has has a proposed AI Bill of Rights for this Florida. Can you say whether or not it's important for states uh control the situation or federal government? And I and while you say that executive orders do not [snorts] override state laws, statutes do. Is that right?
▶ 2:48:18Uh, that is correct under the Supremacy Clause that federal law can over uh overrule state laws that conflict with it. I would say also that in the domain of AI, states have a very real and very acknowledged role in governing uh AI use cases by end users. If and when we see states try to regulate beyond their own borders. However, that infringes upon our federalist system in which Congress, you all, are tasked with governing interstate matters. Thank you, Mr. Chairman.
▶ 2:48:48I thank the gentleman. I thank the the I hope you noticed the uh extent of the numbers of members on both sides of the aisle that showed great interest in making to leave other committees, et cetera, and be here uh to ask questions. This is something that um we will continue as well. So, thank you, each of you, for for sharing your expertise. And uh we know where to find you if we need more answers.
▶ 2:49:19I now would recognize the ranking member for his closing remarks. Thank you, Mr. Chairman. I would firstly ask ask unanimous consent for a um report from the Economic Policy Institute entitled Unbalanced Labor Market Power is what makes technology including a uh threatening to workers. The best AI policy is to protect workers to protect workers is boosting their bargaining position. Without objection.
▶ 2:49:44And the other the um article that uh was mentioned a couple of times the risk of AI in schools outweigh the benefits report says it's a National Public Radio uh piece. Without objection. Thank you. Um this Mr. um Chairman, I think um this is one of a great panel for very um very very instructive.
▶ 2:50:07We've heard a lot about what's going on, the need for I think we've heard one one need that we need some standardized oversight of what um programs are We've heard the importance of asking the right questions. The problem is if you don't have some independent analysis, you don't know what to do with the How's your um program on on privacy?
▶ 2:50:37Oh, it's great. Now, what do you know? Um we need some independence and also like [snorts] we have with the registered apprenticeships the certification of programs with a registered apprenticeship if you have a registered apprenticeship that you completed in um Washington, D.C. and Washington state, they know what you can do.
▶ 2:51:03Uh we need to make sure that certifications in AI have the same [snorts] kind of So, there's a lot uh we've learned today and a lot we need to uh continue to working on particularly with the um uh sinister opportunities for discrimination that are way below the radar that So, thank you, Mr. Chairman, and I yield I thank the gentleman. And yes, uh today is very clear uh dangers, yes.
▶ 2:51:33Um innovations and expansions of opportunity, yes. Both uh job and process changes and eliminations, yes. Uh job and process enhancements, yes. It's all there.
▶ 2:51:48Um policy uh makers must transparently highlight and expose violations and concerns, but with as light touch I believe as possible to foster Not fall into the same trap we fell in with net neutrality.
▶ 2:52:12Um the objective is is I think um a very bipartisan uh that we have a system we have AI that works for all of us and expands Um with caution but that we don't stand unnecessarily in the way either.
▶ 2:52:36Priorities uh as has been said are literacy literacy and true useful development of what goes on. AI will bring about changes in education and the workforce. It has to if we're going to keep up and lead the world. Uh full stop.
▶ 2:52:57Uh but a brave new world has always brought more benefit than had we not bravely move forward with the dangers the pit potential fitfall pitfalls, but the amazing successes that it can bring. How we encourage education and the business community to partner together in the best use of AI is something that we need to continue thinking about. And do it as quickly as possible.
▶ 2:53:26We don't have time to just wait and waste but need to move forward. The The US will lead in AI or the Chinese Communist Party will. I'm not lying, am I?
▶ 2:53:43Uh hence quality and innovative education is a must that is also fostered by a morality that its freedom that comes from information. I think I that said. And of course, factual information that builds on that. Uh optimism is is what I feel.
▶ 2:54:10Uh AI could help produce more sincere and unified bipartisanship through AI uh clarification of how often we may be on the same side but we're unable to clearly articulate our ideas and concerns. And so, I'm optimistic that AI can even help in the legislative process.
▶ 2:54:34Um and uh move us forward on a track that says America will and our kids will grow in in their education endeavors in different ways, in unusual ways from what has been in the past. But on a trend upward with the freedom that we espouse here in the United States that brings about and achievement.
▶ 2:55:01So, again, thank you so much for uh participating with us. And um with that uh and without objection, there being no further business, the committee stands