Hearings to examine the impact of AI on the workplace.

The Subcommittee on Employment and Workforce Safety, Workplace Safety, will come to order. First, it is an honor and a privilege to lead this subcommittee for a few months. I plan to examine the challenges that American workers face and explore policy solutions to increase opportunity, promote fairness, and support families. I appreciate being able to work with you, ranking member Hickenlooper, and I intend to be very active in the time that we have left uh leading this uh subcommittee together. Before I go any further, I ask unanimous consent for Senator Blunt, Rochester, to wave onto the subcommittee for the purpose of today's hearing. Without objection. Um, so moved. I want to welcome our witnesses this afternoon. We have a distinguished panel of artificial intelligence researchers and workforce development leaders to put the impact of ai on the workforce in perspective. According to Goldman Sachs, ai will drive seven trillion dollars of worldwide investment and economic activity over the next decade. The seven trillion dollar question, though, is who benefits from that? But it's not a question for the future. Employers, workers, and governments are exploring that answer right now. AI promises to raise productivity and reshape work. It can also disrupt how the workforce and the workplace is organized and create a whole lot of anxiety. There's no sh- shortage of doom and gloom predictions about mass displacement of workers. We have to take that that seriously. But we haven't seen it begin to happen yet. Based on all the evidence, three things seem to be happening so far. First, hiring of entry-level workers is somewhat weaker. Second, AI is transforming task rather than replacing jobs. And third, the skills that employers need and how workers learn those skills is rapidly changing. The downturn in entry-level hiring is due to several factors, and AI is just one of them. Some researchers believe the Federal Reserve is partly responsible, and I look forward to delving into that today. Even in the most AI-exposed occupations, we haven't seen much change in employment. But the way workers reallocate their time is changing. In some occupations, things like data entry and simple coding are already mostly automated. But checking AI output is a whole new discipline. Finally, when we consider how workers acquire and use skills, we have to think about employment pipelines. A lot of work workforce training happens on the job. Entry-level workers aren't just hired to do their jobs, they're hired to gain experience and grow into more advanced jobs. Every master craftsman was once an apprentice. Every doctor started out as a resident. And most of our staffers in our offices started out as interns. The real challenge is for graduates and young workers to acquire the skills they need to enter the workforce and get on the first rung of their career ladders. If those opportunities dry up, then some of our young workers can't access them. It will be a big problem in the labor market over the long term. Helping workers master the skills they need to succeed in the AI economy starts with getting better data. Existing labor statistics too often fail to tell us how occupations task are changing, what skills are needed, which jobs will grow and shrink, and how workers move through the labor force because of AI. That is why, that's exactly why I introduced the AI workforce prepare act with Senator Hickenlooper, Senator Husted, and Senator Hassan. The Census and Bureau of Labor Statistics need more expertise to analyze and forecast AI's impacts. They need to share more data with state and local governments and collaborate with private researchers. Billions of dollars of spending on workforce development programs rely on good data but better data is not a goal in itself we need to make better decisions to beat China in the AI race. We need to train and equip American workers in nearly every industry to compete and win in the AI economy. We also need to make sure small and medium-sized businesses don't get left behind while the biggest companies sprint ahead. America will win the AI race with China by maximizing the capabilities of all of our companies and all of our workers. I look forward to hearing our witnesses'

Thank you, Chairman Banks. Uh, and thank you for holding this hearing. I wanna thank our panel for making the time to be here and to provide testimony. Uh, throughout human history, the world has experienced industrial revolutions that changed every sector of our economy. First Industrial Revolution obviously driven by steam power, the second by electricity and mass production, the third by electronics. Through it all we've seen hard-working people striving to achieve the best for themselves and the best for their communities. Every technological development has provided tools for these workers, these these strivers to achieve even more, from the telegraph to the assembly line. We are currently witnessing what some are calling a fourth Industrial Revolution. Technological advances in artificial intelligence and automation are changing how we create, how we communicate, how we work, and how we live. These new tools, like the ones that came before, come with great power, but also great responsibility. Small, medium, and large organizations alike are feeling the transformative effects of AI. There's an ever-growing crop of AI assistants, claiming they are "ready to help you revolutionize your house or your business." But today's businesses wonder "How do I know which tools are worth our investment, and which ones over-promise and under-deliver?" Our governments all want to make use of these promising new technologies. Pacific Northwest National Lab has built AI tools to streamline and accelerate the permitting review process for critical infrastructure. Health and Human Services used AI to identify polio outbreaks. And the Colorado Department of Labor and Employment is using AI to improve its customer service experience. But no one feels this revolution more acutely than our workers. Today's worker must navigate a dizzying maze of systems that are increasingly automated. Job applications, interviews, human resource systems, uh Each of these uh entry-level nodes roles are are shifting and often driven by AI. New AI-driven tools are being implemented across sectors with oftentimes little to no training on how to use them. We need to ensure that workers are getting the training they need at every level, whether they're searching for their first job or transitioning to to a new role in the same field or looking to switch industries altogether. Today's workers have got to ask themselves, how am I gonna keep up? Or even better, strive and thrive in this AI ecosystem. That's what's driving this hearing today. As we see AI impacting the workforce in all these different ways, we need to work together on making sure we get the right path forward. How do we help employers, educators, and workers prepare for what's next? We envision a fully connected ecosystem where all these entities are communicating with each other among each other sharing proven practices, forecasting needs, and developing strategies for overcoming these obstacles together. We're working on a bill to encourage just that. Um, our bill brings together regional partnerships of educators, industry and intermediaries to provide credentials. uh apprenticeships, other job, uh other jobs and training uh that use AI. Because who knows better than workers and the businesses and educators in a region that i- in any given region uh than the workers, businesses, and edu- eduators in that region. Uh we've also been working on a bill with Chairman Banks, as he mentioned, and Senators Haassen and Houston called the AI Workforce Prepare Act, which would direct the Department of Labor to Gather gather and analyze data to better inform us exactly how AI is impacting the labor market. Informed policy starts with good data. You can't say that often enough. Too often we don't see it in this building. And this work is gonna provide essential insights into job transitions, layoffs, uh and future worker needs. This is a monumental task ahead of us in this this fourth industrial revolution. The capabilities of current technology are well beyond previous generations' wildest dreams. We are confident that with the guidance provided by people like all of our witnesses here today we can forge a path in the right direction. We need to look at what the research is telling us about how the workforce is shifting and how we can accommodate that and give them a leg up. We need to partner with folks on the ground in our states who are innovating with AI and with workers and employers. We need to shape worker-forward policies to set striving Americans up for success. This is a moment in history when everyone's effort matters. The decisions that we make now are gonna ensure that any child born today continues to have the highest probability of a healthy and happy life of any person born in the history of mankind. Thank you all for your testimony and for taking part in our hearing today, and I yield back to the chair.

Thank you, Ranking Member Hickenlooper. I would now like to welcome our first and only panel. First we have Mr. Ken Clark, the President and CEO of EmployND, the Workforce Board for Indianapolis and Marion County, Indiana. Mr. Clark brings an impressive co- combination of expertise in workforce development, technology and public administration. Mister Clark, you are now recognized for five minutes to to to deliver opening remarks.

Thank you, Chairman Banks, Senator Hickenlooper, and members of the subcommittee. Thank you so much for the opportunity to testify today. Uh, my name is Ken Clark, I serve as President and CEO of EmployND, the workforce development board serving Indianapolis and Marion County in Indiana. Uh, every day we work with employers trying to fill jobs. uh educators preparing students for careers and job seekers navigating a changing labor market that gives us a practical view of how artificial intelligence is changing work today i've had the opportunity to see artificial intelligence from two very different perspectives in my career before leading employee indy i served as chief information officer for the city of indianapolis one of the nation's largest cities helping all three branches of government uh implement new technology uh in managing a seven thousand employee workforce Um Today, I lead our region's workforce development board, uh helping employers find talent and workers prepare for the future of work. In my current role, I have the opportunity to see how new technologies affect employee or employers trying to fill jobs and workers trying to build careers. Those two perspectives have shaped one fundamental belief. Artificial intelligence is not primarily a technology story, it is a workforce story first. Because technology has never been, America's competitive advantage by itself, our workforce has. And that belief has led me to three conclusions, and I'll briefly outline those for you. Uh, first, AI is reshaping far more jobs than it's replacing. Much of the public conversation is focused on mass unemployment. That's not what our employers are telling us. Artificial intelligence is becoming another workplace tool, much like the internet and email. It is changing how work is performed, not eliminating the need for When the new IU Health Hospital, being built now in Indianapolis, opens, AI will help summarize patient information for clinical review for a nurse. Um, the goal isn't to replace nurses, we need more nurses, very clearly, uh, is to give them more time with patients and less time in front of a computer. This is AI augmenting workers, not replacing them. Second, I believe the greatest long-term workforce risk is not widespread unemployment. it's the disruption of career pathways that workers rely on to gain experience and continually develop new skills throughout their careers. Every experienced professional was once an entry-level employee. Today's junior software developer becomes tomorrow's senior engineer. Uh, today's apprentice becomes tomorrow's master trades person, not to steal from you, chairman. Uh, today's uh, uh, if artificial intelligence significantly reduces opportunities for workers to gain those early career experiences, we don't simply lose entry-level jobs, we weaken the talent pipeline employers depend upon to develop tomorrow's workforce. Preparing workers for an ai economy means preserving career-connected learning, internships, apprenticeships, other pathways that develop judgment, communication, human skills that ai cannot replace. Increasingly, employers are looking for workers who can work alongside ai, evaluate ai-generated information, and exercise judgment that only people can provide. That means preparing for AI isn't just about a pathway into the workforce, it's about helping today's workforce continuously learn and adapt as occupations will evolve. And my third conclusion, local workforce boards will play a critical role in helping communities adapt. Every regional economy is different. We're made up of different industries and different talent. Uh, and so that is our own DNA within each region. Um, the AI challenges facing advanced manufacturing looks different from healthcare. Health care is different from logistics. Logistics is different from hospitality. That means workforce solutions must also be very local. Uh, workforce boards sit at the intersection of employers, educators, economic development, and training. We translate employer demand into training strategies, national trends into local action, and emerging technologies into practical workforce solutions. Local workforce boards are only as effective as the information they have, and that is why we support the AI workforce prepare act. Winning the AI race isn't about isn't just about building better technology. It's about building a workforce that's ready to use it. Preparing for AI begins with understanding how it is changing work. Better labor market intelligence, forecasting, and workforce data will help communities prepare workers before disruption becomes displacement. And I'll close with this. America has never succeeded by fearing technological change. We've succeeded by harnessing it. artificial intelligence will not determine the future of work people will if we equip employers educators workforce organizations and workers with better information and tools ai can strengthen america's workforce rather than weaken it I thank you and look forward to your questions

thank you for that next we have miss miss carol rogers the director of the indiana business research research center at the greatest university in america and the returning national champions indiana university's kelly school of business miss rogers is an expert at making complex data useful for real world decisions she serves on boards of several national organizations focus on economic research and labor market information miss rogers you are now recognized for five minutes to to deliver your opening statement

thank you so much chairman banks ranking member hickenlooper and members of the subcommittee and it's a beautiful day out there. So thank you for being here to listen to us. Uh, I was actually thrilled uh to have this opportunity to testify today on the impact of AI on the workforce. It is a real issue. Uh, there are some knowns, many more unknowns, uh, some unknown knowns. Uh, and we could go on, but I'd have to remember the movie that came from. Uh, I do come here as a subject matter, uh, from that great university, uh, in Indiana, uh, but my views don't necessarily reflect those of my employer, the university. So I wanna just make that clear. Uh, as Director of the Center, and thank you so much for giving those remarks about my experience uh chairman banks uh cause i can cut out a little bit on on uh describing why i'm i'm here uh but certainly the thing we have been doing since nineteen twenty five when the university uh trustees of indiana university created our center was to help people in indiana understand our population our economies our workforces, and I say plural because we have many, uh, to, uh, provide them with evidence-based data and research. So, getting on to the, uh, impact on the workforce, uh, generative and agentic AI, I like to, to kind of differentiate a little bit. Uh, AI generically is many things, uh, but certainly Those two have evolved very quickly in the four years since ChatGPT's public release to the world. Uh, nearly twenty percent of US firms today, and actually not today, it was February nineteenth, uh, in a survey by the US Census Bureau. Uh, nearly twenty percent, uh, report using AI in some business function. Uh, Harvard's project on workforce has surveys. looking at the uptake of ai uh particularly generative ai and they're finding that sixty-two percent of adults between the ages of eighteen to sixty-four uh which is kind of our prime training and education workforce uh are using generative ai and workers in that population are reporting spending about six and a half percent of their work hours a week using it. That's, that's kind of hard information. It's useful information. We'd like to see more like that. My center recently published, uh, findings comparing high and low AI-exposed occupations in our state, meaning that they're, they have tasks in those jobs, that some proportion of them could be done by uh a genetic AI. So when we say exposed, in some ways you could say it's they're exposed to the risk of AI, in another way you could look at it as exposed to an opportunity with AI. Like many academicians I uh talk with one hand and then say on the other hand, so you all have to forgive me, but we see both. Uh, Notably, we found that job postings in those exposed occupations actually shrank. Postings for AI exposed jobs, and many of those were in technology, uh, shrank since twenty twenty-two, uh, versus thirty-five percent decline in those less exposed, the jobs that require human touch as opposed to artificial. touch. Uh, wages in those exposed occupations, though, grew faster, uh, but, at the same time, those jobs lost their employment advantage. And think about some of the things that we've seen over the last several months, about college graduates not finding employment as quickly as they used to. Think about jobs, uh, or occupations that people have, uh, where they don't have that two or three percent below unemployment that we used to have, people with bachelor's degrees, master's degrees. However, taking this together, what I've described, and it's not enough, uh, taking it together, it looks more like mass job elimination rather than, or it looks less like mass job elimination, more like ma- market-based, pricing and reallocation uh this bill matters uh i'm running out of time so i wanna say very clearly that i support the bill and its goal to collect data to have better data and my plea is that we create standards for the definition so that we compare apples to apples oh i actually have forty five uh more seconds here

no you've gone over forty five

oh i've done Thank you very much.

Uh, thank you very much. Next we have Doctor Leah Palagashvili. I practiced that all week and I probably still messed it up. Senior Research Fellow and Director of the Labor Policy Project at the Mercatus Center at George Mason University. She is a labor economist whose research focuses on independent workers and how labor markets evolve. Thank you for being with us. Doctor, you have five minutes uh for your opening remarks.

Good a- good afternoon, Chairman Banks, Ranking Member Hickenlooper, and members of the subcommittee. It is an honor to testify before you. My name is Leah Palagosvili. I'm a Labor Economist and Director of the Labor Policy Project at the Mercator Center at George Mason University, where my research examines the changing nature of work and how policy can help build a resilient labor market. Let me start with the most fundamental point. The early evidence does not yet show broad ai driven employment loss much of the public debate asks whether ai will take jobs but economists think of jobs as a bundle of tasks so automating one task doesn't mean the whole job disappears ai can automate some tasks augment others raise productivity and change how firms organize production recent studies find little noticeable change in employment unemployment earnings or hours at ai adopting workplaces according to a census bureau survey about ninety five percent of ai using firms report no change in total employment that does not mean nothing is happening it means the adjustment is more complicated than a simple story of mass displacement the clearest evidence of potential concern is concentrated among young workers entering highly exposed occupations recent research finds employment declining for young workers in the most ai without a matching rise in unemployment, a pattern more consistent with slower hiring than with layoffs. This can be a serious concern. However, the evidence is still mixed. According to research from the Economic Innovation Group, hiring in many highly exposed occupations actually began declining in the spring of twenty twenty two. So, six to eight months before ChatGP ChatGPT even existed. Employment outcomes have also weakened among young workers with little ai exposure at all including teenagers working mostly in food service and retail the the broader pattern points towards macroeconomic factors as much as ai so the right conclusion is not that ai is irrelevant it is that the picture is still mixed and to the extent employment effects exist so far they look like a slower hiring pipeline for younger workers not layoffs This brings this brings me to my second main point about measurement. Federal labor market statistics are valuable, but they were built for an earlier economy. Household surveys can tell us whether someone is employed and what occupation they hold, but not how AI is changing tasks within the job. Business surveys can tell us whether a firm reports using AI, and in some cases whether it says AI affected employment. But those responses are still too coarse and are not routinely linked to hiring, separations, earnings, or occupational composition. Even the basic adoption rate depends heavily on how the question is worded. For example, when the Census Bureau broadened one survey's AI question last November, reported use jumped from ten to seventeen percent overnight. That was a change in measurement, not a month of sudden economic transformation. The AI Workforce Prepare Act is a constructive response to that problem, It adds AI questions to several major federal surveys and creates a voluntary channel for AI developers to share adoption data with statistical agencies. As Congress considers further reforms, I would emphasize three things in particular on measurement. Uh, one, linking business surveys' answers to actual hiring and wage outcomes, so we can compare AI using and non-AI using firms directly. Two, adding occupational detail to the wage records that nearly every employer already reports, so we can see which specific jobs are growing or shrinking. And third, building a timely measure of new solo independent business activity. The last point matters because of what my own research has found. Since early twenty twenty four, new business applications with no signs of hiring um near-term hiring, have risen twenty seven percent in AI exposed sectors while essentially uh saying flat and less exposed sectors. Solo self-employment shows the same pattern, up twenty percent in the most AI exposed occupations, unchanged in the least exposed. These results are descriptive, not causal, but they suggest that in some knowledge intensive fields, AI may be lowering the cost of working independently, rather than reducing employment overall. And this makes sense, if AI helps you do what's what once required a team, you may decide to go out. on your own rather than needing an employer to provide that team for you. And to summarize, the evidence doesn't yet show broad job loss, our data systems can't fully see what's happening, and independent work is growing in ways that matter for labor policy. Let me conclude by saying the central challenge is not to predict a single labor market future, it is to build institutions and measurement systems that are resilient across several possible futures. Thank you for the opportunity to testify and I look forward to your questions.

Thank you. I would like to yield uh to ranking member Hickenlooper to introduce our next two guest. Thank you, Mister Chair. Um first let me introduce uh Mister Jonathan Liebert. Um he's joining us today from Colorado Springs uh where he serves as the CEO of the Better Business Bureau of of Southern Colorado. I think um also working as the uh at least the acting CEO of the whole statewide Better Business Bureau as well.

Um, Mister Liebert created the better, the BBB AI Hub to increase AI literacy um and capability, as well as governance and responsible use. Um, Doctor Justin Heck is the Senior Director of Research in Data Production at Opportunity at Work, a national nonprofit building a skills-first uh labor market. Doctor Heck has spent over a decade conducting quantitative research at the intersection of labor market analysis and public policy. thank you both for being here and participating. Mister Liebert, you're recognized for five minutes.

Thank you. Thank you, Chairman Banks and working member Higgenlooper and members of the subcommittee. Uh, appreciate the opportunity to testify today on the impact of artificial intelligence on America's workforce. Again, my name is Jonathan Liebert and I am the CEO of the Better Business Bureau of Southern Colorado. Uh, I'm also the past chair of the Colorado Workforce Development Board, and I am part of my local workforce board as well. And my perspective on artificial intelligence is is neither pro- ai nor anti-ai i'm cautiously optimistic um i've seen in ways that it can absolutely magnify productivity strengthen small businesses and expand access to expertise i've seen it help workers become more capable i've also seen a people can enter insensitive information into these public systems accept fabricated information as fact and purchase products that are marked as ai powered without understanding truly what that uh technology does Since two thousand and twenty five, I've trained over five thousand and five hundred business owners, um uh employers, educators and non-profit leaders in Colorado and now in other parts of the country, and talking to them about how AI needs to be used ethically, responsibly and practically. This experience has given me really a front row seat to what's going on inside of the workforce. Uh, I'm there to see the excitement, the fear and a lot of misinformation that's surrounding this technology. Business owners really want to know which tools are a credible source of information for them to be able to use safely and they wanna make sure that they are doing things wisely, not putting their customers at risk. Non-profit leaders wanna know how they can expand their impact and their limited staff resources and level up their organizations. Educators wanna know what to prepare students for, and knowing that jobs are changing fast and employers are looking for different things and what they should put into their curriculum. Workers want to know whether AI will replace them, and they want to know what will help them become more valuable workers to the workforce system. And so what I've learned is that fear often decreases when people receive practical education. The greatest problem I see is not that people are refusing to use AI at the moment, it is often that they trust it too quickly. And so I tell the folks that I teach that AI is not an oracle, AI is an ideation engine. It gives you the ability to create these templates and information, but you have to keep the human in the loop and double-check everything. An oracle is an answer to be believed. uh, an ideation engine helps generate these options and you should challenge these assumptions, it'll help you organize information and produce that first draft. The human must remain responsible for verifying the output, adding additional context, and then absolutely making the final decision of how it's implemented. At the Better Business Bureau, we created the BBBAI hub to help close the gap between the speed of technological change and the ability of people and organizations to respond quickly. BBB is interested in making sure that Trust and ethics are applied when using this technology. And again, as our mission and vision state, we're interested in marketplace trust. If buyers and sellers don't trust one another because of the way the technology is being used then that has economic implications for all of us. In collaboration with our local workforce center, the National Alliance for Workforce Boards, Chamber of Commerce, City of Colorado Springs, and partners across business, education, government, we created this AI hub to really focus on a couple of different priorities. Uh, one is AI literacy. The second is workforce preparation, AI governance, and absolutely marketplace trust. AI literacy means more than learning how to use a simple chatbot. It means understanding what AI can and cannot do, how to verify its output, how to protect sensitive information, and how to recognize misinformation and bias, and more specifically, when that human oversight is required. AI governance means establishing clear rules before problems occur. Organization should have robust AI policies and determine which tool should be used by their teams and develop robust ai literacy plans to help them keep up with these very very quickly changing systems one of the largest issues that ai has is that most people cannot conceptualize how to use it since it is moving so quickly additionally we all need to get used to the fact that how we work is changing and we need to learn how to review ai generated work in a productive way we must still think critically as we engage with ai and remain accountable for when ai affects an important decision Marketplace trust is also extremely essential. Small business owners and nonprofits are being approached by a growing number of companies and uh consultants that are saying they have these AI powered claims or AI can fix these things. Most smaller organizations do not have legal teams, cyber security experts, nor technology procurement departments. And so they need practical guidance, guidance on how to evaluate vendors, uh how to update their data practices, how to look into these performance claims, and look at security contract terms and human accountability. The workforce is changing, and in many cases, AI is going to help individuals automate these tasks and do jobs that they never could before. But again, we have to be careful that we're not fully automating people out of the first rung of this career ladder, and then wonder why workers can't reach the second. So again, I think Congress can help by doing a lot of practical AI literacy, obviously enacting different types of acts will help to create more data and information for all of us to use. And again, I would say this, that America does not need to choose between innovation and responsibility. We can pursue both at the same time. And in the future, the work will not be predetermined by an algorithm, it will be shaped by all of us and the choices that we make, as we lead our workers into an educated system where we protect, make sure people have access to the tools, and make sure that we hold businesses and consumers accountable. So again, thank you for the opportunity to testify, I look forward to your questions.

Thank you.

Doctor Heck.

Chairman Banks, Ranking Member Heckenlooper, members of the subcommittee. Thank you for the opportunity to testify. My name is Justin Heck, and I lead research in data production at Opportunity at Work, a non-profit focused on building a skills-based labor market. Public debate about AI in work has focused almost entirely on whether jobs will be augmented or automated. Far less attention has gone to how AI will reshape economic mobility. That's what I wanna talk about today. Workers build skills in one job. and use them in the next. In aggregate, these moves form pathways. We analyzed a hundred and thirty million job transitions over ten years, and found three kinds of jobs. Entry-level origin jobs, higher-wage destination jobs, and in between, the stepping stones we call gateway jobs. Sixty million Americans work in gateway jobs, and sixty-two percent of them are stars, workers skilled through alternative routes. who've built their skills through work, military service, apprenticeships, training programs, or community college, rather than a bachelor's degree. Consider a customer service representative. Workers reach that job from roles like receptionist or cashier. And on the job, handling frustrated customers, finding solutions under pressure. They build critical thinking, speaking, and persuasion that prepare them for the better paying HR or sales job that comes next. AI can break that stepping stone function in two key ways. The first is straightforward. There may be fewer of these jobs. Among the ten largest occupations with high AI task exposure, half are gateway roles. And the Bureau of Labor Statistics projects declining demand for all five. That cuts both ways. A cashier or receptionist loses the next rung up, and employers may find fewer experienced workers ready to move up into management or senior roles. The second is less obvious, and it may matter more. AI can hollow out a gateway job from the inside by eliminating what a worker learns from doing it. If that customer service rep spends their time overseeing a chatbot, rather than handling the hard calls directly, they lose the opportunity to develop the skills that would have carried them to the next rung. The learning that happens on the job often happens through the hard parts. AI is exceptionally good at removing the hard parts. This can happen without anyone deciding to do it. Learning has historically been the by-product of getting work done, not the objective. A company that automates the demanding parts of jobs may see short-term efficiency gains while inadvertently dismantling the pipeline that produces its own future supervisors and managers. Displacement debates miss this. A worker whose job is eliminated knows it immediately. A worker whose learning has been quietly removed from the job may not find out for years until they apply for the next job and discover they no longer qualify. None of this is inevitable. AI can augment work rather than hollow it out, but whether AI narrows pathways or strengthen them depends on how the technology is deployed and on the choices before this subcommittee. Congress can act in three ways. First, build a skills-based labor market. When pathways get disrupted, workers need their capabilities recognized wherever those capabilities were built. The Federal Jobs for Stars Act would embed skills-based hiring in federal employment, while the skills-based federal contracting act would do the same for federal contractors. Second, treat investment in workers the way we treat investment in machines. Employers can more easily write off an AI system, robot, or server than they can the cost of training workers to use the same technology. If we want employers to choose augmentation, the tax code should not be telling them the machine is the cheaper bet. Third, fund what works, and measure what's actually happening. The Better Jobs Through Evidence and Innovation Act would fund proven programs. Reauthorizing WIOA with adequate funding would give the system we already have, the resources we need to meet this moment. And the AI Workforce Prepare Act is a meaningful step toward the data we need, including data to see how AI is reshaping jobs, and the pathways that run through them. I'll close with this. AI presents enormous opportunities. But as it reshapes jobs, it will reshape the skills that workers build and the pathways available to them. If we do this without intention, we lose more than jobs. We lose the routes people use to move up and the talent employers need at the other end. That's a choice, not a forecast. Thank you. I look forward to your questions.

Thank you to each of you. Uh, good opening statements. We will now move on.

Alright, thank you, Chairman. I'll let you all answer the same question here, and I'm looking for a thirty to forty-five second answer so all of you can do it. I'm a big fan of community colleges, technical colleges. So let's just assume that there's a significant number of people, in probably data entry type of people that are gonna lose their jobs to AI, and they're gonna need to go back and get some some more training. They've already got a bachelor's degree, right? I don't think they need to get necessarily another, but what advice would you have for community colleges, technical colleges to help quickly retrain these people and pivot back in and, mister Clark, where to start with you and and, you know, two or three, two or three things that that you think would be helpful, maybe there's something you assess what we shouldn't do is okay with me as well.

Yeah, uh, thank you, Senator, happy to answer. Um, for community college, the key here is workforce Pell is gonna be a huge part of the answer.

What a good idea.

The investments that are gonna be made in workforce Pell in Indiana, now worked out workforce Pell and our have the opportunity for our community college, both ivy tech and vincent's university in indiana be able to skill up that's gonna be a big part the university has to respond that community college has to respond with the right training

yup

uh and so they're still working on that but that would be my number one

great great answer too by the way m- miss rogers

thank you senator uh and good question you know how do we retrain them when they're displaced over the past uh fifty to seventy five years congress has enacted a significant legislation to help displaced workers uh find jobs uh i am glad that we're gonna be looking at that again and i i would argue that each state uh already has employment projections part of the bill is to enhance those projections uh i'm one of that argues we can enhance it with uh more information around wage records. Uh but I would encourage people in my state, for example, to look to the state of Indiana's uh already existing top jobs platform that helps show people within regions close to where they live which jobs are in highest demand and have a high wage. But we've set it up, and there's a group of people who have worked on this, so that you can kind of ratchet it based on the capabilities of the person, right? So they may only have a high school degree, look for the jobs that are high wage, high demand for those skills, and then work with them

Right.

to increase it.

Thank you. Doctor, go ahead. I'm not even gonna try your last name.

Don't worry. I I will mention what my co-panelists uh uh as jerry s over here mentioned which was in my written testimony i think there is a very interesting overlooked low hanging fruit solution uh reform which is that um congress could uh restore um tax neutrality between investing in machines and investing in people uh since the tax code currently makes it easier to deduct the cost of um automation equipment or any other capital investment

yeah

of training workers to use that equipment or that capital. And I I think that is uh one way to help incentivize um firms and um a market-based solution way to help incentivize firms to also train in workers as well.

Thank you. Mister Liebert.

Thank you, Senator, and so I agree with with uh much of that's been set up here, so I'll add a couple of things that haven't been mentioned yet. One, I think you're bringing up a very good question because we need to utilize the existing infrastructure that's already in place. And so I think that that's absolutely critical. So utilizing the community college system, chamber of commerce or bbb or any of those organizations are already there i do think that there in order to talk about this whole thing that's already in place of the re-skilling upskilling certification credentialing is these career pathways and so getting more data to help us understand which pathway they're going into and which type of task they need to be taught i think is key but the other thing i'd add to the community college system is just uh making sure that we're doing the train the trainer so this ai literacy component is absolutely necessary to make sure that we're getting it in the hands of the teachers the educators the skills that they'll need to kind of have these conversations with the students. And then also making sure that there's some good correlation and connection between um colleges and universities as well as the business community to make sure that we're accessing the skills they need. This stuff changes so, so fast. Uh, one of the, the partnerships that we're working on right now with the Better Business Bureau and the Colorado Community College System of Colorado is is making sure that we're in line with some of the classes and training and formal education that they have. And then coming to the BBBAI hub for these kind of quick hit lessons that are just, we're uh able to be a little bit more quick and fluid, but we're partnering together to make sure we're providing necessary, ever-changing kind of skills and curriculum to uh the folks in Colorado.

Great. Doctor Hicks, could you just a brief answer please, sorry.

No worries. We, I I think we need to focus on the skills that folks already have, rather than assuming they need to start from square one. Most workers aren't going to go become an AI engineer. We should take seriously the domain expertise that they already have.

OK.

Um, and we and we should make sure we bring employers to the table, so we know that that we're training workers for jobs that exist.

Thanks. Yeah Mike, a quick question for the Chairman and ranking member to to contemplate in my, is they they tell me we're gonna lose a lot of entry-level positions, accounting, um, look how you know, how do you get get experienced accountants then if we fire all the entry-level accountants and attorneys then I'd be interested to, we'd that we keep pursuing that answer too.

Thank you. Ranking member Hickenlooper.

And I'll hold maintain my place and yield to Senator Baldwin for questions.

Thank you. Uh, if I get to it, I have a very similar question to the one that you just uh, some of the panelists, but I have a couple first. Um, so, uh, Wisconsin is all too familiar with the disruption that comes when Washington makes the wrong decision or fails, uh, to plan ahead. So as we consider policy and regulation on AI, we have a responsibility to working men and women to get this right. Um, something we have heard today is that we must have access to the best data to understand how ai is changing uh the workplace and impacting workers so um doctor hacken i i think i wanna ask miss rogers the same question thank you both for being here today one of the policy recommendations that you alluded to in your testimony is improving data collection what gaps exist in our current data and what should we be doing differently to better measure and more precisely measure ai's impact on workers Start with Doctor Haq.

Thank you so much for the question. I I I think the panel has already brought up a lot of good examples of how we can improve our data systems. I I'll highlight two in particular. Uh, the first is thinking about how workers are actually moving between jobs. Much of that data we only see two to three years after it occurs. We need really current indicators to see how workers are making these moves, and we need to be able to see um how their jobs are changing, what they're spending their time on, how AI is coming into the picture. Um, I I would also add that um, linking to existing systems is a really great solution. So when we think about the great data that exists at the States, how can we take that data, connect it in ways where it's more useful for everyone? Thank you.

Great. Thank you, Miss Rogers.

Thank you very much, Senator. three accounts really uh have stood out to me uh in uh working uh toward understanding the bill. First, our occupational classification. Uh, our definitional structure really don't accommodate uh understanding what the job is there uh you know you can't just say it's an ai job or it has ai use in it uh we need to work on that and come up with uh definitions that are already out there uh among many institutions certainly academic institutions but also uh companies themselves like meta and Facebook and Amazon, how are they defining the use of AI in their jobs. Uh second, uh and to to echo your point about this, uh states already hold valuable data about workers uh as well as students, and we have those data linked through the uh state longitudinal data system or sleds as some of us call it. Uh and third, uh we need to figure out what the firms themselves are doing in in adopting ai governing ai uh implementing it into different portions of the firm is not a one size fits all situation and i think it's really important for us to know how they're allocating work uh based on how they're implementing ai