Senate Health, Education, Labor, and Pensions Subcommittee on Education and the American Family (Murkowski): Hearings to examine the future of K-12 education in the age of artificial intelligence.

The Senate HELP Committee on Education and the American Family will come to order. Before coming to the Senate, I spent decades working with young people as a teacher, a coach, and mentor. I've been in classrooms, locker rooms, communities where the goal was always the same: help students reach their potential and Prepare for the future. That's why today's hearing is so important. Artificial intelligence is changing the world our kids are growing up in, and whether you like it or not, AI is going to be part of their education, their careers, and their daily lives.

The question isn't whether it's coming. AI is here. The real question is whether we're going to help students use it the right way As a coach, I always told my players that talent is important, but you have to have skin in the game. The same is true in education. Technology can be a valuable tool, but it will never replace a great teacher, a strong mentor, or the hard work that learning requires.

AI has the potential to help students learn at their own pace, give teachers more time to focus on teaching, and open doors for students who may need extra support. Those are opportunities we should take very, very seriously. But we also need to be honest about the challenges. How do we protect student privacy? How do we make sure AI is given accurate information?

How do we prevent students from becoming dependent on technology instead of developing critical thinking skills? How do we ensure that rural America isn't left behind? Most importantly, how do we make sure AI strengthens education instead of weakening it? Our job isn't to chase every new trend. Our job is to figure out what works, what doesn't work, and what's best for students and parents and teachers.

The students sitting in the classrooms today will enter a workforce that looks very different. Thank you. From the one many of us grew up in. We have the responsibility to make sure they're prepared not just to use AI, but to compete, innovate, and lead in a world where AI will be everywhere. Two weeks ago, my ranking member and I led a letter to the Government Accountability Office asking them to begin an investigation into the effects of AI on K-12 education.

I'm proud to report that they have responded and plan to start work on that study very soon. Today I look forward to hearing from our witnesses about how we can encourage innovation while protecting students and preserving the values that make education successful in the first place. So thank you to our witnesses for being here today. I look forward to your testimony. And with that, I'd— Thank you.

I defer to my ranking member, Senator Blunt Rochester, for her opening statement. Thank you, Mr. Chairman. And thank you so much to you and your team for partnering with us on a hearing that I think is vitally important for our country. And I also want to thank the witnesses for being here as well. Before I get into my formal remarks, I do want to begin by acknowledging today's announcement that the Department of Education is attempting to follow through on Secretary McMahon's plan to shut down the department by transferring some of the core functions, like today, to the function of protecting students' civil rights and students with disabilities to other agencies without the approval of Congress.

And I— while I strongly oppose those actions, I'll continue to call on our chairman, Chairman Cassidy, to bring Secretary McMahon before our committee to discuss these moves, to also be held accountable, and really to make sure that our American education system is strong and protects our children. The hearing today, I think, is one that is very timely. I mean, we find ourselves in a moment of great uncertainty. You could call it a tale of two virtual realities. Some people are afraid we're going to have a Star Trek moment.

Some people are afraid it's going to be the Terminator. [FOREIGN LANGUAGE] We want to make sure that it is in the best interest of our students and our educators. And artificial intelligence seems to be everywhere. Our schools, our classrooms, and our students' homes are no exception. Yet as AI reaches all parts of our lives, there are unanswered questions about its impacts, and the stakes are getting even higher alongside them.

That is why I'm so glad that we're holding the hearing today and why I was so pleased to work with the chairman and Senator Kaine to request that investigation by the Government Accountability Office on the current state of AI in K-12 education. I believe we should be gathering as much accurate information as possible in order to make informed decisions that will keep our students safe and help them succeed, as well as support our educators. I come to this topic both as a ranking member of the subcommittee, but also as a grandmother. I have a granddaughter who is now 3 years old and will be educated and is being educated in an AI age. And that means she's growing up with a new way, new things than that we had when she graduates from high school in 2039.

I want her to be able to be creative and have critical thinking skills to succeed. I want her to be interested in learning and understand that not all questions can be answered instantaneously. I want her to have the understanding of what it means to go to work, to start a business, and what careers are out there in this rapidly changing economy. And alongside my dreams for my granddaughter and for all students in this country, I also have concerns. The impact of AI will have on cognitive development and learning is not well understood, and some experts are sounding the alarm that cognitive skills are being negatively affected.

The guidelines about what safe and ethical AI should look like in the classroom are not well defined and vary greatly from place to place. Yet reports show that 84% of high school students are using generative AI for schoolwork. Similarly, reports find that 85% of teachers are using AI in their work. I know teachers are looking for professional development opportunities and clear guidance to support safe and effective use of AI. Schools, districts, states across this country are trying to craft responsible guidance and identify AI tools that will support student learning while protecting their civil rights, mental health, and data privacy.

My home state of Delaware is doing amazing work on this front, and I look forward to hearing from Secretary Cindy Martin about how our state is leading on this issue. Importantly, I think all of us here want to know what we as Congress can do to support the work. And I hope that this hearing helps us better understand the challenges and the opportunities that AI presents in education. With that, I yield back. Thank you, Senator.

Now we'll get to our witnesses. First, I'd like to introduce Josh Jones, proud Alabamian and a leader in AI space. Mr. Jones, uh, is the CEO of QuantHub, a Birmingham-based company that builds AI and data literacy curriculum for schools, universities, and employers. He has partnered with my state's Department of Education and a number of businesses throughout Alabama to prepare for the age of AI. Mr. Jones' partnership with our State Department of Education is a huge reason why Alabama students have some of the best numbers nationwide in enrollment in data science courses and data literacy.

Quan Hub is reaching every field, training educators administrators, and our future workforce. He also created QuantHub's Data Scholars Program that has placed 100 high school undergraduate students into paid data and AI internships. QuantHub is creating opportunities for students to earn industrial recognized credentials in data and AI literacy before they even get their diploma. Mr. Jones has a proven track record of tackling the world of AI headfirst. Thank you.

Mr. Jones, your opening statement, please. Chairman Tuberville, Ranking Member Blunt Rochester, members of the subcommittee, thank you for this opportunity to testify today. I am Joshua Jones, CEO at Quant Hub. And as the senator said, we're an Alabama-based company that builds AI and data literacy curriculum for schools, universities, and employers. And our mission is really to empower talent with the digital agility to thrive in this evolving world.

I'm also honored to be joined today by our State Superintendent of Education, Dr. Eric Mackey, as well as my colleague, Dr. Veronique Brown. For the past 4 years, we partnered with the Alabama Department of Education on what I believe is one of the most ambitious statewide efforts in the country to prepare K-12 and postsecondary learners for an AI-driven economy. Now, Alabama is not a state most people would associate with leading-edge tech education. And that's part of why what's happened there really matters. In its recent state of the field report, the National Data Science for Everyone Initiative ranked Alabama 3rd in the nation in student enrollment in data science and related courses, only behind California and Virginia.

And through our efforts, we placed over 100 high school students through a pilot program into paid AI and data internships throughout the state. Teachers and administrators have earned thousands of professional development hours Again, in the National Data Science for Everyone report, we were ranked number 1 in the country in terms of the most professional development hours. Students are earning industry-recognized credentials in data and AI literacy before they leave high school. As Dr. Eric Mackey has put it, integrating this kind of training into the state's educational framework is about preparing students for the future of work while keeping Alabama competitive in a rapidly evolving digital economy. I'd like to share today 3 lessons from this work that I believe are directly relevant to federal policy.

First, we must refresh our standards of curriculum at the speed of technology. The traditional curriculum refresh rate of is typically 5 to 10 years. Yet in the age of AI, that's generations. At Quant Hub, we've built a human-in-the-loop system that uses AI agents for curriculum generation, but it always validates with instructional designers, faculty, and subject matter experts before anything ever reaches a student. We've reduced curriculum development hours from roughly 200 hours to create 1 learner hour to about 10 hours of instructional validation.

That's a 20x improvement without compromising rigor, efficacy, or accuracy. Now, in the interest of sharing these advancements with the education community, we published our process in the peer-reviewed journal, the Journal of Applied Instructional Design. Federal investments in workforce curriculum should reward speed and update cadence not just initial development, learner hours, or page counts. Second, the human side is where this work succeeds or fails. For years, data and analytics courses have lived on the fringes as obscure electives.

The breakthrough in Alabama was not a better course. It was building modular content that maps directly to existing state standards in math, science, social studies, and CTE. So a teacher doesn't have to add a class, they integrate a unit into one that they already teach. Now, that requires heavy investment in teacher training, instructional coaching, and integration support. Curriculum is a small piece of the puzzle.

Change management is that critical linchpin. Third, you can't build a workforce pipeline from inside a school building alone. Our Data Scholars Internship Program works because we built bridges to employers, civic leaders, nonprofits, higher education partners, and we match students to real paid roles. We didn't put these interns in jobs pouring coffee, but we made sure that the corporations were actually identifying upfront meaningful jobs related to data and AI so that these students could build real-world skills. And in many cases, they're getting job offers as soon as they complete their internship.

Our Data Scholars program works. We've seen more than 50 employers host our scholars now. And in work-based learning, I would suggest that recruiting employers at that scale is typically the hardest part of the program, which makes the number itself a strong signal that this talent is in real demand. It also gives us a continuous feedback loop that keeps the curriculum aligned with what these employers actually need. The microcredentials that our scholars earn are directly aligned with Alabama's Talent Triad, a talent marketplace where a credential isn't just a certificate, it plugs into the system that the employers actually use to find workers.

Now, the pilot for the internship program was actually funded by a state innovation agency called Innovate Alabama, and it's been recognized as a model for rural talent retention. Now, this matters enormously in a state, in a country where small towns are losing young people to larger metros. 100 Placements on limited funding tells us that this model scales. Now, with dedicated federal support, particularly through Perkins V and existing workforce pathways, it can be replicated in any state. In closing, I'd respectfully offer 3 recommendations for consideration.

1, Fund curriculum infrastructure that can be updated continuously, not just developed once. Refresh rates should happen in semesters, not quadrennium or decades. 2, Weight teacher training and classroom integration at least as heavily as content creation in federal workforce education grants. And 3, expand support for paid work-based learning in AI and data fields starting in high school. Thank you.

The Alabama model is replicable. It's not expensive relative to what it returns. And it works because we treated this as a workforce problem, an educator problem, and a curriculum problem, not just a technology problem. Thank you. Thank you very much.

Our next witness is Ms. Erin Moat, founder of— co-founder of InnovateEDU.

An organization that prioritizes innovation in education and empowering students' use of technology in the classroom. She is also the founder of Project Unicorn and the leader of the EdSafe AI Alliance. Ms. Mote has a storied career of adapting technology for students and continuously finding solutions to put our nation's students ahead of the game. She brings extensive experience in the education technology space and securing our national digital infrastructure. Through EdSafe AI Alliance, she has partnered with multiple state and school districts to bridge the gap of AI with students and teachers.

While superintendents and administrators start panicking at how AI will change education, Ms. Moat gave real solutions to adopt AI and not run from it. Ms. Moat. Thank you. Chairman Tuberville, Ranking Member Blunt Rochester, and distinguished members of the HELP Committee, thank you for convening this critical hearing.

My name is Erin Mote, and I'm the CEO of InnovateEDU, a nonprofit organization dedicated to ensuring technology serves as a powerful tool for access and opportunity for all students. I come before this committee not only as a charter school founder, but also as a technologist who spent more than two decades working at the intersection of technology policy and systemic change. Yet perhaps my most important role is a deeply personal one. I'm the mother of two school-aged children. Each day I see firsthand how these technologies impact real kids at the kitchen table and in classrooms, which fuels my urgency to ensure that our national digital infrastructure is secure, safe, and intentionally built to serve the next generation.

To navigate this landscape, we must reject a pervasive false dichotomy. Safety is not the counterpolarity to innovation. Safety is not a drag on progress. Rather, safety is a necessary condition for the effective use of AI in education, the very guardrail that makes meaningful, sustainable innovation possible. Without it, we mistake rapid tool adoption for progress.

Generative AI is an arrival technology like the internet or electricity. It is reshaping how students, educators, and our society live, work, and relate to each other. Data shows that roughly 85% of teachers and 84% of students are already using generative AI in their classrooms. Thank you. But this transition did not follow a linear path of institutional procurement or risk evaluation.

As a result, school districts and families are experiencing severe operational fatigue. More than half of schools have failed to provide any professional development on the safe use of AI. Our schools urgently need federal leadership, structured support, and dedicated funding. A consumer large language model is engineered for platform retention and user satisfaction. To keep users engaged, it defaults to a dangerous form of algorithmic sycophancy.

It does not teach, it indulges. It validates incorrect premises and hand-delivers answers, often bypassing the productive struggle required for cognitive development. Mm-hmm. Learning is an active process. True purpose-built edtech scaffolds learning with a student zone of proximal development, much like the AI-driven math accelerators currently being utilized in Alabama to bridge foundational gaps for learners.

When we substitute this with consumer-grade systems, we trigger what Wharton researchers call cognitive surrender. When human psychology meets this increasingly agreeable tech, independent reason is often relinquished. Alarming data shows that when a chatbot intentionally gave incorrect answers, users accepted the error 80% of the time. We are actively placing yet spots in front of neurologically vulnerable youth, conditioning a generation to surrender critical discernment. Moreover, today the average school district accesses nearly 3,000 —digital tools a year, creating a fragmented, unmanageable landscape for teachers, families, and students.

We must transition to evidence-based purchasing anchored in 5 edtech quality indicators: ensuring tools are safe, interoperable, usable, inclusive, and evidence-based. Meeting a baseline level of evidence—Tier 4 within the federal ESSA framework—must be a non-negotiable ticket to enter America's classrooms. When less than half of purpose-built edtech tools meet this standard, only 2% of consumer tools do, we must shift our focus to a new ROI, return on instruction. We also must replace the blunt, outdated metric of screen time with screen value. Counting minutes does not tell us if an interaction is meaningful.

For instance, screen-based reading is 6 to 8 times less effective for reading comprehension than physical books. Yet for decoding, a skill essential to the science of reading that requires 1,000 at-bats to achieve fluency, purpose-built AI excels. Furthermore, blanket bans or hourly limits can disproportionately harm the many children with disabilities or who, who rely on digital tools for mandated accommodations. We cannot rely on digital avoidance. Congress must ensure that AI is built for our children's safety, learning, and healthy development, while also building tech literacy and digital responsibility in our students.

To protect our students without stifling innovation or widening the digital divide, Congress must approach a waterfall approach to safety. This means applying heavy-handed safety controls to high-risk applications at the top, such as public consumer platforms and relationship-stimulating chatbots. Thank you. While easing the regulatory burden on lower-risk, purpose-driven educational tools. Finally, the federal government must enact immediate structural policy changes, reconstitute and fully staff the Office of Education Technology, establish a joint interdisciplinary AI in education federal research agenda and infrastructure through the NSF, IES, and NIH, protect E-RIGHT and fund non-consensual intimate imagery enforcement agencies like the FTC.

We cannot afford a passive wait-and-see approach. We must design digital regulatory ecosystems at the federal and state level that aggressively target genuine harms while keeping human values, ethics, and critical judgment at the center of the American classroom. Thank you, and I look forward to your questions.

And it's my honor to introduce, uh, Secretary Cindy Martin, who is the Secretary of Education in Delaware. She has more than 35 years of experience in education leadership, policy, and classroom instruction. She holds a Master of Arts in Teaching and Learning from the University of California, San Diego, and a Bachelor of Science in Education from the University of Wisconsin-La Crosse. Secretary Martin began her career as a teacher. In the years since, she has been a leader at every level.

[FOREIGN LANGUAGE] serving as a principal, a superintendent of San Diego Unified School District, and as Deputy Secretary of the United States Department of Education, where she oversaw K-12 initiatives, including the Office of Education Technology. She has done remarkable work, from greatly improving literacy outcomes at the school district level to leading efforts to close opportunity gaps at the federal level. Thank you. I'm grateful that she has brought her knowledge and expertise to Delaware, where she is now leading on education reform in our state. Secretary, Madam Secretary, thank you for being with us today, and I will turn it over to you for your testimony.

Thank you very much, Chairman Tuberville and Ranking Member Blunt Rochester. It's an honor to be here today before the members of this subcommittee. To address this important topic. As mentioned, I've spent more than 3 decades as a teacher, a principal, a superintendent, and Deputy Secretary at the U.S. Department of Education. And today I am proudly serving as Delaware Secretary of Education, where our focus is to start with students and then build for impact.

Across every one of those roles where I've served, one thing has always remained true. Our responsibility is not to prepare students for the world that we grew up in, but to prepare students for the world that they are entering. That's why in Delaware, we are approaching artificial intelligence with both urgency and care that's required. I want to start with something that a Delaware student recently told us. The student said, AI should be a collaborator and a tool, not something that does the work for us.

And I think that is exactly right. This conversation is not really about technology. This is about learning. It's about critical thinking and helping students to develop the very skills that they need to enter into the world where AI is going to be a part of nearly every single profession. The reality is, as we've heard, AI is already in our schools.

Students are using it. Families are asking about it. And our educators are left navigating it. So the question is not whether AI is going to impact education. The real question is whether we will shape its use thoughtfully, responsibly.

And our answer in Delaware is yes. Our work supports a goal in Delaware's strategic plan that 75% of our districts and charter schools will use AI tools to support educators and students while they maintain full compliance with data security, privacy, and the human-in-the-loop requirements. Under Governor Matt Myers' leadership, Delaware has chosen to lead. We are developing statewide guidance. We are investing in AI literacy and building what we believe is one of the nation's first statewide AI assurance efforts focused on K-12 education.

In 2024, Delaware released statewide guidance to support effective use of generative AI. Thank you. In schools. That guidance is grounded in 4 priorities: protect student data, ensure ethical use, promote equity, and enhance teaching and learning. These priorities are consistent with the national guidance and developer guides and toolkits that were published by the Federal Office of EdTech.

We're also listening closely to our students. One student told us, "We have to make sure students still develop critical thinking skills, not just let the AI do all the thinking for us." straight from our students. Another student emphasized that AI-generated information has to be checked, has to be checked because systems can make mistakes and hallucinate. Many of the students raised concerns about data privacy and the need for adults to protect student information. See, what strikes me is that students aren't asking us to slow down innovation.

They're asking us to lead and lead responsibly. They're asking us to do what good educators have always done. Mm-hmm. Teach them how to think, not what to think. That's why Delaware is taking the next step through our AI Assurance Lab.

The goal of our AI Assurance Lab is not to slow innovation. It is to ensure that innovation is worthy of the trust that students, families, and educators place in our schools. The lab creates this trusted, transparent process for evaluating AI tools before schools adopt them at scale. Delaware recently completed a statewide scan of AI use, and what we found was encouraging, but it reinforced the need for this independent review. Most AI use today is focused on helping educators with productivity and planning.

That's a great data point. If we use AI and use it well, teachers can get time back and do the human work that only teachers can do. We— districts and charters still have unanswered questions, though. They want to know about privacy protections, integrations, and how tools actually perform in real education settings. That is why our assurance lab does not simply review the vendor claims.

Vision has to come before vendor. Trained educators, administrators, technical experts, and DOE staff are testing tools used in our Delaware classroom scenarios. We're evaluating tools against national frameworks and our own state guidance, and we're looking to make sure our tools are protecting students and whether the outputs are accurate and appropriate and whether there is bias showing up or not and how accessibility is working to lead towards better teaching, learning, and outcomes. This is practical implementation work. It's not technology for technology's sake.

Because a safe tool is not automatically a useful tool. We need to help educators and families understand the tools and earn greater public trust. And we need to support our educators, not sideline them. We want students to develop judgment, creativity, and critical thinking. Mm-hmm.

That no technology will ever replace. When we keep students at the center, lead with strong guardrails and shared responsibility, and judge AI by outcomes rather than hype, I believe AI can expand opportunity while preserving what matters most about education: human relationships, human judgment, and human learning. Thank you, and I look forward to your questions. Thank you very much. Now we'll start a round of questioning.

We'll start with Senator Banks. Thank you, Mr. Chairman. To you and the ranking member, what an important subject and topic to host this subcommittee on. And to all of you for being here, I'm encouraged by the audience, the— those of you who have come to participate and listen to this and such an important conversation. Ms. Moat, like you, I have 3 daughters in public schools, high school, middle school.

They're in a great public school district with great teachers, administrators, and a school board. I think they're trying to figure all of this out, but I don't think they have yet. I don't think any school yet has figured out where we're going with this, and I think we all agree we have to be flexible. I do think, Secretary Martin, that the answers to a lot of these questions are going to come from Indiana and Delaware and the states much more than they're going to come from bureaucrats. Anybody in Washington, D.C.

So the partnership that we have with you and those at the state level, I think, is really important and why this, this subcommittee hearing is so important today. I wonder, just to start with, if all of you could be just a little more specific about what are the— what are the AI skills today that our schools should be teaching students? And if you could just sum that up in a succinct way. And then maybe you could talk a little bit as well. I'll defer to each of you.

How do we measure that? How do we measure that our students are being taught effectively in this new age and adapting to these AI skills? But if you, Secretary, if you could just start with like what are the skills that teachers in Delaware are teaching or that they should be teaching to kids about AI and the use of it? That is adding to a more constructive and productive learning environment? Thank you for that question.

And the skill base is for the students and also for the educators. So there's skill— AI literacy that we speak about. First, the educators have to have a skill set. And then I think you're asking specifically about student skills. How do we use the tools that are part of the teaching and learning that's happening in the classroom?

Teachers have their state standards.— and things that they're required to teach and content that must be learned by the end of a year. And as they're teaching, how can a tool be used to enhance the critical thinking that students are doing? So as they're creating work, as they're working together, as they're doing research, are these tools allowing them to personalize their learning, to make it more related to their own background and their own interests, and make the learning more meaningful for them? So the skill base ends up being how do you pick up this tool and what am I going to use it for? You could go to— I'll use, since it's a tool analogy, you could go to any— I won't say the name of a store, but any tool store, buy a tool.

It's how are you going to use this to create this end-of-course paper that you're going to write? The skills have to do with literacy and they have to do with math and they have to be connected to science as well. And each of these tools end up being something that if a student student knows how to use it, they'll be able to produce a stronger product, and they'll be able to evaluate critically the information that the tool is producing. They can't just accept the answer as it's, as it's being produced. That's great.

Ms. Mo, can you talk more about what the, the specific skills the teachers should be teaching kids? Thank you so much for the question. I think we need to be thinking about AI literacy as a foundational literacy across all subjects. That it's important that we're able to interrogate the inputs and the outputs of this technology in history and in biology and in computer science. And so when I think about what I want to cultivate in my own children, I want them to be critical consumers of this technology.

I want them to grow the skill of discernment. And frankly, that means working in multifunctional teams with humans, but also with AI as a teammate. Mm-hmm. And increasingly, I think that's what workplaces are demanding, that AI is a tool and a teammate in the work that they're doing to discuss, to discern, and to be in discourse with each other. And Mr. Jones, I mean, is there, is there a way that schools can measure that?

So absolutely. And I might add from a skill perspective, if I could, just a couple of the quick skills that we offer. So the foundations of AI, what is it, where did it come from? The basic ethics of when should I be using AI, when is that okay? Tool selection.

You know, in the early days, particularly of generative AI, we had one vendor that was the name everyone knew. But now you have so many different leading vendors understanding their strengths and weaknesses. When would I use one tool versus the other based on the job or the task at hand? I would suggest— we don't call it this, but a healthy cynicism around not just assuming that everything that AI tells you is correct. And so how do we teach these students to understand that this is basically an informed opinion that you need to really judge with your own research?

So to the— and we go well beyond that. But to your question around how do we test this, something that we've been a little bit hesitant, or I should say very hesitant to do, is expose learners directly to large language models due to the predictive nature of these. So everything that we do in teaching, we use AI on the back end to accelerate curriculum creation. But it always has the backstop of validation by subject matter experts. So students are actually getting a fixed version of that educational material.

So when we talk about actually testing them, how do you test when if you put a prompt into an AI, you know, a large language model 3 times and you get 4 different answers, how do you test that from a rigor standpoint and an accreditation standpoint? So what we've done is we've created controlled environments with simulations where we leverage really the best of both of those technologies. But as we're teaching them prompting, we're not just teaching them the tactical say this word or this, because that's changing so fast. Almost by the time you publish it, it's going to be, you know, chances are out of date. So how do you back up and teach those critical thinking skills?

So what should I be using? You know, what should I be using this for? What are the strengths and weaknesses? Those sorts of things. And those are things that we can definitely test.

And so we do those and what we refer to as our summative assessments. Thank you. My time has expired.

I am so excited about this conversation, and apparently so is America, because the line is out the door and around the corner to get into this room. We have a lot of young people in the audience as well. Raise your hand if you're under 40. The whole audience just— okay, this is what happens when hallucination of the, But I have to say, so I don't know if this is a large learning model glitch or what, but back here. But I will say it really demonstrates just how timely this hearing is, just how important these issues are.

As you were talking, Mr. Jones, I was thinking recently, University of Delaware has an AI literacy test that you can take. So I took the test myself. To see how much do I know? Do I know about the ethical issues? Do I know about how to protect myself?

I've even had these models argue a prompt. I put it in the same prompt into all three, like maybe, I won't name names either, but a couple of different ones and get different answers. And then I have to say critically, well, what's right? And so to me, I'm glad that you mentioned not relying just on what spit out, what's, you know, comes out, because what we put in is what comes out. And also change management.

I think we've done a lot of things in the past using technology. It was change management that was necessary to make people prepared and accepting and be able to use appropriately. My first question I want to ask is to Secretary Martin. I would love to hear some of the things that you think have been the most exciting things that you're seeing right now in Delaware as it pertains to AI in education. And then what have been the biggest challenges that you've faced?

And lastly, in that question, how can we at the federal level support it? Because my colleague just shared a lot of incredible things are happening at the state level, almost as laboratories to help us to figure out what we need to do next. But are there things that only we could do at the federal level? So one, share anything that is exciting happening in Delaware. Two, some challenges.

And three, what is it that we in Congress should be focused on? And that question I will ask of all three. Okay. Thank you. Thank you for the question.

I am very excited about what I see in Delaware. Having had the opportunity to lead at the federal level and the guidance and toolkits and materials that we were able to publish, I'm now in a position leading a state to be able to implement that strong guidance that came out of the Office of Educational Technology that included playbooks, that included vendor guidance, that included teacher K-12 guidance. And what I see in Delaware is so much optimism, but that comes with a lot of responsibility. So what's exciting to me is this tremendous potential and that Delaware has been working on this before I came as the Secretary of Education. In their generative AI guidance, and they see the promise in being able to reduce administrative burden for teachers, being able to give teachers more time with students to take some things off a teacher's already very busy plate.

I see the promise. I'm excited about opportunities for accessibility for students with disability having more access to learning through some tools that help them access. If it's an auditory deficit or visual processing, these tools are helping. That's a lot of promise there. Mm-hmm.

And the ability to give feedback as students are learning real immediate feedback, not waiting till the end of the year to know if somebody's learned and what you're going to do about it, but instantaneous feedback with teaching and learning in the classroom, always with the teacher, right with the students. So all of that excitement happens because it's strengthening the accessibility and improving operational tasks that teachers have to do. That's exciting. Anything that has to do with differentiation, translation, all of that is good. Mm-hmm.

The challenges are real, though. The challenges are that this technology is evolving faster than systems are made to adapt. And what I see are challenges around privacy, security, procurement, instructional use, the bias that is built into the models, the governance that needs to happen with limited capacity and the district's ability to be able to create this guidance quickly and be able to act on it so that they can enable innovation. We do not want to be stopped by fear. We want to be able to move forward.

So being able to— make complex decisions and be supported with federal guidance at the state level so district leaders, elected school board members, and superintendents have clear evidence-based paths forward. Mr. Chairman, if you would indulge for that second question, I'm just curious what can and should we be doing at this level? Ms. Mote. Thank you for that question.

One thing that we need to understand is what works for whom and under what conditions. Right now, there are currently no high-quality causal studies on the long-term effects of AI on student learning, equity, or social-emotional development. The federal government is uniquely positioned to help us understand both what we know and what we don't know so that we can empower states and educators, families and communities to make decisions about AI with knowledge. And facts. And so there's a place where the federal government, by working together to understand the mental health aspects, the education aspects, and the learning aspects, can really help lead communities and states with evidence to guide their decisions.

Thank you. And Mr. Jones? I, I would suggest at Quant Hub, we're really focused on the refresh rates of curriculum. We're in an environment where the technology is advancing on an exponential exponential curve. And what we were seeing now with these AI models, for example, building other AI models, the pace of change is increasing faster and faster.

And so from an educational standpoint, how do we deliver curriculum that is built knowing that this curriculum is only going to last for 6 months and we have to have, you know, so, so that refresh rate, I would suggest when you're thinking about funding different, both from all sorts of teacher training, educator training, all the way down to the students, everything that you're funding, you have to realize that that time factor is something that we've never seen before, and it's going to continue to increase. I'm going to yield back my time, but I have— if we go into another round of questions, I have a lot more for the record. But for all the young people who are here or watching, please reach out to our offices. Send us your thoughts. You can do it through our websites.

You can mail it to us, however is best for you. But we want to hear from you too. So thank you so much for being here. And I yield back. Yeah, they can use AI, right?

They can use AI. Senator. Thank you, Mr. Chairman and Ranking Member. Appreciate you having this committee. I was actually— this committee hearing, I was actually involved in AI at the state government level prior to becoming a U.S. senator, in which before ChatGPT was even introduced on November 30th, 2022, we had already used an AI tool to recommend 5 million words for elimination out of the state's regulatory code of 17 million words using an AI tool with human collaboration.

And then, so this is so new 'cause we're, most people's consciousness didn't really understand what AI was in their lives even though it existed until we had this release in November of 2022. And since that time, we created, I know in Ohio we created an AI toolkit for trusted resources for parents, educators, students to look at. And you're correct, Mr. Jones, this stuff needs to be updated constantly because things change so quickly. But in conversations I've had with educators and experience I've had as a parent, educators tell me that have incorporated this in the classroom that they've used it— and they feel more knowledgeable about the progress of students on an individual level than they've ever felt before. I talked to one language arts teacher who told me that, that over the course of the day she has something like 140 students, and she knows exactly where each one of them are in their development of their language skills because she can follow it along, and then she can reach those students on an individualized basis to help them learn and to reach them where they are.

So I see a lot of heads nodding. It seems like this. And I also believe it can be an egalitarian technology in the sense that every student will have access to a tutor. And these AI tools get better. They're better at it.

Yeah. If you were gonna ask me to help you with your pre-calc homework, I wouldn't be very good at it. But my AI tools can. And they can help your children have a tutor 24/7, 365. So those are the virtuous things.

Of course, we don't know about what the harm may be, just like with social media tools that we know have become harmful in the lives of children. So it's a constant process, and creating an awareness around this is very important. And so, so I want to ask you, and I'll start with you, Mr. Jones, and we'll go right down the line. Just quickly, so parents that are concerned about this, what should they be asking their local school leadership about AI to feel comfortable that it's being used appropriately in their schools? I think one of the great disservices educators do in many cases is they don't address it.

They stick their head in the sand, they hope it'll go away. And so the worst thing you can do is just have a syllabus or, or not clearly defined rules. So talking about many, particularly in university settings, in a class, green light, yellow light, red light situation. We're going to use AI in these situations, but here's a situation where you should not use it because in some cases it can help you do things like research much better. But in others, you should, you should set it aside, use critical thinking skills.

And so really having your educators just addressing that upfront, not saying it's, it's the great evil, we're going to stay away from it, but also not using it everywhere, but really being methodical about where are we going to incorporate it? Into our classes, how are we going to use it? Start by teaching those ethical skills up front and just being thoughtful in your approach to it. Okay, Ms. Moat.

Thank you. Parents are fearful that their students' data is being used in a way that is not protected. And so the most important thing that educators can do is to actually say, this is what this tool is being used for. This is how I'm using it in my human-centered relationship with your student, and really address that fear straight on. At SafeAI, Common Sense Media, and the National Parents Union, actually earlier this year released 10 questions for parents at parent-teacher conferences to have those conversations with their educator about how that educator is using that tool in the classroom and what it looks like for that experience.

Transparency is the most important thing that we can be doing right now in our education ecosystems. And Secretary Martin. Thank you. I noticed in your two examples you totally kept the human in the loop, which is exactly what we're talking about. That teacher that you referenced is a teacher who wanted to use a tool to get to know her students more so she could help them in their individualized, personalized learning.

That is the promise. Parents can ask the same hard questions that a teacher is asking. Is this helping me teach better? A parent is asking, is this helping me— helping my student learn more? I think healthy concern is good.

The questions that my colleague just mentioned, they should be in plain language so parents can walk into a parent-teacher conference, can walk into a school and ask the key questions. What data— what tool are you using? What data on my student is being used? How is that data being stored? What biases may or may not exist?

How are adults monitoring these tools, or are these tools on autopilot with no adult ever actually looking at it? And most importantly, how is this tool improving learning outcomes for students? If it's not, don't use it. And then quickly, if I could, Mr. Chairman, I know that the ranking member and I have the RAISE Act that we've sponsored that would require states to set a statewide standard for the use of AI in the classroom. We want it to be at the state level, not at the federal level.

How important, Secretary Martin, do you believe it is for states to have a standard that, that then local schools can use as a reference point for all of those issues so that their safety, security, using it in the most appropriate way? I think it's critically important, and this is speaking serving at the federal level, a district level, and now at a state level. At the district level, the level of expertise that is needed, not every district, especially rural districts—. Sure. Might not have.

Maybe large urbans might have a whole team that can procure properly and put the guardrails in place, but not everybody. So the state can play a critical role, which is what our federal guidance suggested. And now I have the opportunity to lead in Delaware. Actually have a student saying, "We're so glad that you have some guardrails because we want our teachers using these tools. We just want them using it the right way." So when the state can put that kind of guidance out so not every district has to recreate the safety and the guardrails and the standards, these are known quantities that the state can lead.

Thank you. A state standard, trusted resources, constantly updating it, that's the starting point that every state should be at. Thank you, Mr. Chairman.

I think we're on round 2 of my questions. And so I'd like to kind of pick off where— pick up where Ms. Boat left off on privacy and data breaches. And in 2025, 52% of U.S. school districts experienced a cybersecurity incident. The year prior, a data breach breach at the company PowerSchool exposed the data of 62 million students and 9.5 million teachers, the largest breach of children's educational data in U.S. history. So my first question is to you, Secretary Martin.

How do you deal with a challenge like this at a state level? It leads to the other question that I just answered about how important it is to have the state— statewide safety guardrails. In Delaware, we work with DTI to make sure this is a state-level approach, that the security and the cybersecurity in particular, which is also work that I helped to lead at the federal level, there is a knowledge base out here about what to do to protect this. And this can happen with strong state guidance, but more than guidance, it's actually enforcement to make sure every district, every tool— I use the analogy of a superhighway. Mm-hmm.

It's got to be safe. When you build a highway, you build the guardrails. You don't get to come into Delaware and bring a product into our schools unless it meets this minimum safety requirement, just like you don't put a bicycle or a skateboard on a freeway. I'm using that analogy because people get it. For AI in education, making sure that these tools from a cyber perspective and a student privacy perspective are completely guardrailed before they even hit the road.

That is possible to do. The knowledge base exists, and that's about state leadership. Thank you. And for all of the witnesses, do you think that educational privacy laws cover enough ground, or do you think we need to update them? And I'll start with you, Mr. Jones, if you have an opinion.

If not, we can move to Ms. Moat. I don't have an opinion. Yeah. Ms. Moat. Thank you so much.

Currently, FERPA, COPPA, and a number of privacy laws are doing a good job in educating Mm-hmm. I think the challenge is when we have consumer-grade tools who enter our classrooms and might not be subject to those laws and those restrictions. In particular, I'm especially concerned with chatbots and companions and the use of chatbots and companions by young people where we have seen that there is foreseeable harm, much like social media, the infinite scrolling, the consumer passive, nature. And so when those technologies come into our classrooms without the guidelines and the guardrails for their use, it can put students' mental health at risk. And Secretary Martin, do we need to update any of our privacy laws regarding education?

Because of what my colleague just said about the chatbots and some of the ways that students are interacting with the technology, we have learned —there's access to—the answer is yes. Because there is—these tools are getting more access to more information about our students that we may not even be aware of. This is why our Assurance Lab, and when we're putting our tools through the Assurance Lab, we are uncovering not the vendor claim, but the actual what happens, what information has just been harvested or learned in an interaction between a student and the machine or the tool. That is now living in a system that we do not know has the right protections. There is a way to protect this well.

I will shift to AI and the connection to civil rights. We know that the Office for Civil Rights has seen the lowest number of cases resolved in this past— over the past decade in this year. And then with the news that we just got today from the Secretary of Education, Ms. Mote, what do conversations around civil rights and AI in education look like among advocates and parents and groups? And can you give us some real examples of what is actually happening in the civil rights space? Thank you.

It's important to note that about 70% of parents share the worries of advocacy groups like the National Parents Union, ACLU, Ed Trust, the Center for Democracy and Technology around 4 specific types of harms. Bias in AI detection tools, particularly for grading. Mm-hmm. Discriminatory predictive analysis. So, looking at a tool, and because you say the word y'all, you are put into a lower cognitive category.

Student privacy and surveillance, how that data is being used. And really this fear around a digital divide that our rural students and students in under-resourced schools aren't going to have access to these tools like other well-resourced schools will. Mm-hmm. And so I think there's a number of things that are being done right now. One, AI literacy for parents.

Many of these groups are doing incredible work with schools and state governments to really empower parents. I'll use Florida as an excellent example of a state who's really taking leadership here, thinking about an issue of consequential decision-making. This is thinking about how we keep a human in the loop so that it's not the AI who's making a decision about whether you get a loan, whether you get into college, whether you get financial aid, what's the result of your insurance claim, but that there's always a human that's reviewing that decision. And Florida has really led that work around an AI civil rights bill and as well as consequential decision-making legislation. Thank you.

And I yield back and maybe we'll come for round 3. Thank you very much. Senator Armstrong. Thank you. And first of all, thank you all very much for being here.

And really fascinating topic. And it's great to have your expertise on this. I want to go back to the round 1 question from Senator Blunt Rochester, because I think, you know, for this committee, understanding exactly what we can do from a federal perspective is really critical. And so this is a fascinating topic, and we could talk for days on best practices and everything else. And I know that a lot of states and a lot of the local communities are taking on— a lot of groups like Mr. Jones' group you've been involved with are doing a great job of showing leadership on the topic as well.

But my question really goes back to that. Like, what I heard— the response I heard a while ago from the question of what can we do at a federal level? What I think was research, maybe something on best practices, and perhaps maybe something on critical safety standards. But I just want to make sure that as we come through this listening session, we're hearing really specifically, like, what should we be doing at the federal level on this? So, Mr. Jones, I'll start with you, please.

Thank you. And I may actually tie back to the question I passed on a minute ago on policy. Just thinking about your question, one of the things as an entrepreneur, when we go into other states, whether it's privacy policies or AI regulations, we have to be mindful of maybe 50 different sets of privacy policies or 50 different sets of regulatory practices. As a small business, that creates a significant burden that instead of innovating or improving our product, we're spending money on legal or other efforts. So I would— not to take too strong of a stance on the issue, but I would suggest anything that you can do to really allow for entrepreneurship, allow for opportunities for exploration in this area would be the first thing.

And the second thing, going back to this idea of refresh rates, don't think in terms of, hey, we just did something, so we'll check back in 5 years and see how that's going, but looking at what does a semester-by-semester approach look to refreshing, analyzing what's working, what's yielding real outcomes, and moving at a much faster cadence. Because I can guarantee you our competitors overseas are doing that. Yeah, so I'm, I'm hearing that as don't think you can keep up with it, with law, um, particularly at the congressional level. And but, but do put the money into research. I, I really— I don't— I'm not wanting to put words in your mouth.

I'm just trying to make sure I've got it really clear. I would suggest in terms of whether you're passing policies or whether you're funding different initiatives and opportunities, shorten your cadence on when you look for results and when you reanalyze and you spend more in those areas. I would say as an entrepreneur, innovation happens in small companies many times. And so don't get distracted just by the hyperscalers and the multi-billion and trillion-dollar companies, but realize a lot can be accomplished through smaller businesses as well. Strong believer in that.

Thank you. Ms. Moat. I want to name some existing federal programs that I think we need to make sure we're investing in. So first, we need to prepare educators. They are the frontline deliverers of both literacy, but also how our students are going to be interacting with this technology in a human-centered way.

And so fully funding Title II and Title IV dollars is critical here. Those are professional development dollars. That can be used for educators to get this necessary training. Like I said, there's barely half of schools have done professional development for our educators at this point since 2022. I would also name that this, uh, Congress must protect E-Rate technology money.

This is vital internet connectivity dollars that support rural communities in making sure that that young people have access to the internet in schools. This is critical money in order to ensure that we are not widening the digital divide. And then I wanna go back to the question about cybersecurity. The FCC also has a cybersecurity pilot right now that is so oversubscribed that they can't keep up with this work. That means our education system needs more resources to protect our students' and educators' data.

Congress needs to make that investment. Great, thank you. Good answer, thank you. Secretary Martin. I will double-click on that, the investment in the educators.

A tool is no good if the person that's using the tool doesn't know how to use it. So the funding streams that she talked, Title IV and professional development matters. This is implementation support that you can— let's say we get the safest tool and it has all the right guardrails. If somebody doesn't know how to use it to actually improve instruction to the skills question that was asked in the first round, we need real implementation support. We need real time for educators to learn together using real student cases with the very students that are in front of them.

And I'm not talking about an— like an asynchronous model that comes with the tool that you just go on 2 hours and think you've been trained how to do it. The implementation can't be an afterthought. I think that professional learning can't just be a tiny line item after you purchase a big platform. Teacher agency and voice and time to learn to use the tools and the safety, the risks and the benefits of the tool needs to happen. So we invest in implementation, we invest in coaching, collaboration, and then we build trust that way.

And I think that professional learning ends up being a really important investment that can be made. We also need federal support for research, as you mentioned, and evidence building. And voluntary best practice frameworks so we're not recreating— every district has to recreate the best practice frameworks, every state has to do it. There is good knowledge base out there and we can actually not take away local decision-making, help states and districts make informed decisions without replacing the local decision-making, but get that knowledge base out there so states can lead. Great.

Thank you very much. I yield back. Thank you, Senator Armstrong. Senator Kaine. Thank you, Mr.

Chair, and Senator Rochester for having this really important hearing. Maybe just to my colleagues, you know, think who was the most— and everybody in the room— think who was the most influential teacher you ever had in your life, you know, and think about that person.

There's competition in my mind for that person. So, I had a high school teacher who, because I was an introvert, thought it would be good for me to participate in the extracurricular of speech and debate because that might make me more confident in communicating. It was not part of the class, it was not anything I was graded on, it was somebody who noticed something about me and thought this extracurricular could be helpful. It was the shop teacher in middle school, and I was not that great at, shop teacher, at least not on the woodshop side, who used to love to go camping and take kids with him because he didn't have kids of his own, and who would invite shop students, whether they were good or bad, getting an A or an F, like, hey, you want to go camping this weekend? I want to do it.

That was not what my family did. It turned me into a lifelong outdoors person that I wouldn't have been without Joe McKenzie inviting me to go camping. But again, that wasn't part of the class, and it's not something that AI could do. Thank you. AI wouldn't have noticed that here's somebody, though introverted, might gain confidence in doing some public speaking.

AI wouldn't have noticed that here's somebody who might really like to do outdoor stuff. And so AI, I think, has great promise, but the foundational relationship between a child and a teacher is not something that AI is going to recreate, not— Recreate. Not something that AI will substitute for. Now, I think it was Ms. Moat, in your testimony you write AI can help by, quote, offloading administrative functions to technology. I think that's a real good point.

If we can, you know, ease some of the paperwork and other burdens that teachers or administrators or school board members have to deal with, I think that can be really helpful. So there are certainly benefits AI can provide in K-12 classroom setting. My wife's a member of the Richmond Human School Board, and they're grappling with this issue of how to appropriately bring AI into the classroom without expecting it to take over functions that can never be taken over. So here's a question I want to ask maybe all of you, and then I'll have one more follow-up. A lot of school boards are now going to be— consultants will be coming to them like, oh, you got to do something on AI, and we've got the best program in the world, and you know, you should invest a lot of money in us.

What should educators, particularly maybe superintendents and school boards, be looking at as they assess whether to, you know, adopt AI tools or not? Give them some guidance and maybe some warning. Secretary Mark, you want to start first? They are grappling with it because it's very real and there's a lot of pressure to do the right thing. And that's why I mentioned earlier that there needs to be a vision before there's an answer to a vendor who's trying to sell you a product.

What is your district's educational vision? What outcomes are you committed to? How are you supporting your educators? Then you go and find some tools, but you also have to put the guardrails. I mentioned the document that we produced, the Empowering Education Leaders toolkit, and in this toolkit we put 3 clear steps.

It's a roadmap. It's a playbook, if you will. First, mitigate the risk. Second, build a strategy around your instructional core. And third, guide and evaluate the use over time.

And then each of these 3 things has modules in it that— Mm-hmm. Is clear guidance for a school board trying to navigate this. This is published. It's out there. It's not the only one, but it's what we publish at the federal level.

And it actually shows you exactly how to implement this roadmap for safe, ethical, equitable AI. I want to make sure that Ms. Mote and Mr. Jones have a chance to offer advice to superintendents, principals, and school boards about the way to do this right. Thank you. I think we have to ask this very important question. What is this tool's return on instruction?

What will this do for a student? What will this do for an educator? And if it's about productivity or motivation or engagement, does that mean that that educator is gonna have the thing they want the most, time back to make sure that all students are safe, known, and loved? For me, it was Miss Tang in 5th grade, Senator Kaine, who changed the way I thought about myself as a leader. Thank you.

Mr. Jones. So we have developed curriculum specifically teaching administrators and principals, and, and we start before AI. We start on basic data literacy, helping them understand how to really evaluate the data in front of them. And we go all the way to AI use.

And I would say one of the things that's probably most critical that we teach as quickly as possible for those that start using AI, there's a tendency to trust everything that it spits out. That can create serious problems if you're evaluating students, you're evaluating performance, and you're trusting the AI's math. And so, as I mentioned earlier, just this healthy cynicism of evaluating the results is going to be one of those critical skills. And the tool selection— we try not to teach a specific tool, but rather abstract one layer and say, what are the capabilities? How do you understand that it's a marketplace and there are different vendors and competitors, each with strengths and weaknesses?

And chances are you may use one tool for one purpose, another tool for another purpose. So if I were advising directly an administrator, I would say try not to get locked into one system so that your hands are tied, because 12 months from now, that's probably not going to be the best system for you. Let me just say, as I hand it back to the chair, I'm going to ask a question for the record that— to give each of you a chance to share best practices. Like, if you know of a school or a jurisdiction that's doing best practice processes and implementing AI tools in the classroom or training teachers and others to be really good at AI, I'm going to ask you to give us some examples because I think sometimes we talk about the problems and we don't hold up the examples that if others knew about the good examples, they could recreate it or recreate those examples. And the last thing I'll say just on AI and this point about steering you wrong, I never asked my phone to summarize email communications.

But now if I have a chain of emails with somebody, there's an AI summary at the top. And the number of times that the summary is wrong— so-and-so, your friends have agreed that they do want to get together this weekend and go on a hike. No, I can read the emails. They said they didn't want to go and they want to do it next weekend. The number of times that the summary is incorrect has made me quite wary about— and this is non-malicious inaccuracy.

And there's also malicious inaccuracy I'm worried about too. Thanks, Mr. Chairman. Thank you. Mr. Chair.

Thank you. Senator Murphy. Uh, thank you very much, uh, Mr. Chairman. Um, I don't think there's any way to overhype the damage that AI has already done to our kids' ability to succeed. I'm the parent of an incoming 12th grader, an incoming 9th grader, and I've seen already how at scale kids have outsourced critical thinking, have outsourced friendship, have even outsourced moral advice to AI, completely unregulated interactions between machines and young children, something we would've never thought of allowing just decades ago.

All your testimony was compelling, but Ms. Mota, I thought yours was especially compelling given your talk about this idea of cognitive surrender that can happen. Mm-hmm. And let me maybe ask this question about cognitive surrender first. So I like that you're talking about AI as a tool for teachers, not just unleashing it on kids as guinea pigs. But the AI companies have a veneer of social benefit, but an underlying motivation of addiction and profit expansion.

This is from a chilling New York Times exposé this weekend. Regarding the unleashing of deepfakes without any controls on the American public. One observer of the major tech companies said, "These companies would burn everything to the ground as long as they're making a profit." So let's say we're successful in just having this pointed at teachers as a tool. But if the AI companies are only interested in selling more, in having teachers use it more, How do you protect against teachers being sucked into some kind of algorithmic addictive product like kids are? If the intent is not social benefit, but the product being used for more and more and more, how do you make sure that teachers aren't being sucked in like kids are being sucked in?

Thank you for the question. I will wanna give full credit to professors at Wharton who came up with this study around cognitive surrender and this idea that we are offloading tasks and offloading the discernment that Senator Kaine just talked about when he looks at those AI summaries. So there's two things I think that are important, Senator Murphy, when we think about, uh, students versus teachers. The first is what neuroscience tells us. There's a really big difference in putting this technology in front of a brain that is developing and is doing that offloading, frankly, in a way where those neuropathways are still developing.

The difference between Senator Kaine's experience and that of a young person is that Senator Kaine has the ability to read what his friends are saying, knows their preferences, and has those neuropathways really concrete. So I think there is a different set of risk with young people than there is with folks who frankly are over the age of 25. Mm-hmm. If I had the rules, I would say it's rental car rules, 25 and over, 'cause that's really when we know the brain has gelled in terms of those neuropathways. But we also need to invest in professional development for educators.

Here's what we know about actual effective uses of AI, is that when AI amplifies human potential, when the human alone is working with a student, learning can go up about 4.5 percentage points. But when the AI is providing that just-in-time data diagnostic, when it's potentially giving a hint with that judgment, students grow 10 percentage points with the same tool. And so it really is how is AI supporting that human-centered knowledge transfer? That's the type of work we need to be focusing on in education. But it's, I think it might be a little naive to suggest that an adult is magically immune to these incredibly strong brainworm-like algorithms.

I agree with you that they are more immune than children. You know, we require a teacher to go through licensing before they are in a classroom. We require most complicated technologies to go through an approval process before they are handed out to the American public. What do you think about an idea to make sure that AI technology that is being handed to a teacher go through some expert approval process first to make sure that its actual function is to supplement the teacher's instruction rather than its primary function is to try to spin you into higher usage and higher pay rates for the company that owns it? Thank you for that question.

We're developing a learning sciences benchmark actually right now with Arizona State University to measure consumer-grade AI tools. Tools and the impact that they have on learning and cognition to give educators and decision makers an idea of how a tool stacks up. For me, the work that Delaware is doing, Utah is doing, even Vermont is doing in creating sandboxes at the state level to test these tools, to build evidence, and to make that evidence digestible to educators is a huge— Thank you. Huge improvement on how we're going to be able to make sure that all schools have access to information about what works for whom and under what conditions. You were nodding your head, Secretary Martin.

Just wanted to give you a chance to respond. I'm very much nodding my head at this because you're asking to interrupt an algorithmic cycle that the one that can interrupt that is the human. We can actually see it for what it is, realize we're being taken down a rabbit hole, and stop it. But that's at the individual level. You're asking at a systemic level, nationally, statewide, districtwide.

We want that cycle interrupted, and it has to do with what is being demanded as the outcome of public education. What do we want our students leaving knowing and being able to do so a teacher's not going down a rabbit hole that's not producing the outcome of student learning? We want critical thinking. We want kids to be able to design and defend— Mm-hmm. Their point of view and their perspective on something.

No tool is going to take you— you're in some rabbit hole trying to figure out the actual— not the problem that you're trying to solve, which is student learning. I want to know, did students learn? Was I teaching and did students learn? And I appreciate I'm over time, Ms. Chairman, but I imagine that a product, right, that is interested first in return to investors are, for instance, going to point you to the purchase of other products that they make, right, that might have nothing to do with student outcomes but are certainly aligned with the profit return. And so I just want to make sure we really think through what the underlying motivations are of the people who are making the software before we just assume that their motivations are the benefit of our kids.

Thank you, Mr. Chairman. Thank you, Senator Murphy. I just got a few questions here. Mr. Jones, You're uniquely positioned to understand which particular AI skills pay off for students at which grade level. In your opinion, which AI literacy skills should students develop before graduation?

I would say— I would suggest, Mr. Chairman, absolutely, that they should develop what we refer to as AI literacy. These basics of understanding ethics, when to use AI, when not to use AI, tool selection, what the different tools are for. Critical judgment— how do I evaluate the outputs of these tools? What can I trust? What can I not trust?

And then applied AI, beginning to understand how AI is being used in different industries. Those would be the first areas that we would recommend. How can schools prepare students for careers that will increasingly utilize AI? I think one of the interesting questions about workforce development and skill preparation is it's a shifting foundation. I know when I went to school, I was told that I needed to learn long division because I wouldn't be walking around with a calculator in my pocket.

Well, turns out I am, right? So do we still need to learn those skills, or can we fast forward knowing that we have a calculator, a computer to do those things? And so what we would suggest is, you know, going back to this neuroscience, this idea of cognitive difficulty, how do we teach where if it's too easy, students will forget it, if it's too hard, they'll give up and go away, but that desirable difficulty right in the middle where it really challenges them. And I would suggest in terms of workforce development skills, those are rapidly changing. So what we're doing is looking at an industry-by-industry basis and saying, how is AI in Alabama?

How is AI changing agriculture? How are we using that to increase crop yields or for pest control? How is AI changing healthcare? And in each of those, it really is the, the next level beyond that basic AI literacy. Okay.

Becomes industry by industry. And so I think some people, when they respond to, hey, we need to do AI preparation, they want to go launch an AI major or focus on the AI. But really, that's a highly technical field. What we need to be thinking about is whether you're a truck driver or a mechanical engineer or a nurse, how is AI changing your industry? And those skills are going to be— there's those foundational skills that cut across those industries, but they're also going to be very specific skills that we need to prepare our students for.

Thank you. Secretary Martin, in your experience, how can AI enhance rather than replace the role of teachers? It becomes such an important tool when a teacher has something that they can use to reduce administrative burden, which we've talked about, so that they're in real time with their, with their students. A gentleman recently just mentioned around knowing what students have learned. Have they mastered the content?

Where are their gaps in their learning? And have they caught up to the language that they were supposed to be learning? He just gave that example of a teacher friend of his, the senator that just left. That enhances learning when teachers get real-time feedback right away. The question is, when you teach something, how do you know if students have learned it?

Can you get real feedback knowing if students have learned the content? But also how they're interacting with the knowledge, how they're defending their argument, how they're creating argument. And how they're practicing their critical thinking. Thank you, Ms. Moat. For now, the answer may be we don't know yet.

But what do we know when it comes to long-term cognitive impact of the use of this technology? I know we talked about it a little bit. What's your thoughts? I think it's very important to say there's a lot we don't know, Senator Tuberville, and Including in K-12 education, we have no causal studies on long-term impact on social or cognitive development. What we have is a number of studies that are short-term and tool-based, that are about practices, that are about pedagogy, that are about teacher moves.

And both Secretary Martin and Mr. Jones have given good examples of educated hypotheses that we have about how we can build the skills that students need, the cognitive development, the literacy. But we need more focus on research so that we can understand exactly what you said in your opening statement, what works for whom and under what conditions. Ms. Jones, your thoughts on that?

Could you repeat the question? We don't know yet. Okay. But what do we know when it comes to the long-term cognitive impact, use of this technology? You know, as a father of 3 daughters, watching them grow up in high school, college, and middle school, it certainly— I've seen 3 different generations essentially even in that short span because of the way that this has rapidly developed.

And it certainly concerns me the pace at which this, this technology is developing. So I would be remiss if I didn't say, as a father, I am nervous. This is something that we absolutely need to work towards addressing, and I would certainly encourage studies sooner rather than later. Thank you. Ms. Martin, in your experience, how does teacher professional development play into this process?

I think the core argument for this whole discussion today is actually teacher professional development. We want to know what works and what doesn't work. Let's ask the teachers doing the work. They will tell you which tools are not going to improve student learning and outcomes. They will tell you which tools will actually help them teach students so students can learn and think critically.

And professional development so they understand the inherent risks so they don't use the tools in dangerous ways. But you can use these tools to absolutely improve student outcomes when you're working with teachers. This is why we developed the Assurance Lab in Delaware to test the tools in relation to the actual students that we're teaching, the standards that we're required to teach, and are they producing outcomes or is it just a side project over here because it's a shiny new toy? Mm-hmm. You have to test the vendor claims.

The vendors will claim, 10 reading scores improvement or whatever the claim is going to be, let's have our teachers actually test the tools with students and the curriculum that they're teaching and they will tell you if learning is enhanced and outcomes are improving. Give them the time and the space and the professional development to know how to test those tools well. Thank you. I've got one more question here. And we'll start with Mr. Jones.

Everybody can answer this if you'd like. Will AI narrow achievement gaps or widen them? I think that's a very pertinent question, and I would suggest we face a risk of a greater digital divide than that which we saw with access to internet. So I think it's incredibly important that we make sure that as these technologies develop, that we ensure all of our students, our teachers, and administrators have access to evaluate these tools, to use them in an ethical manner. But I would suggest that those that are using it are seeing outsized gains in productivity and efficiency.

We see in the workforce higher demand for those that come in AI fluent. So it's definitely something that's going to be impactful. And I would say it's, it's of the utmost importance that we do educate our learners really for, for this changing environment. Thank you. I think that is the critical question of this hearing, which is what What future do we want for our young people, for our schools, and for public education?

And I believe that if we fund our schools' digital defense, we fund teacher training, and we build a research infrastructure, we can protect that sacred spark of human potential that Senator Blunt Rochester talked about in her 3-year-old granddaughter. I see in my Robert and Claire, and really secure a prosperous future for all American children. Thank you. It's up to us to bend the arc. Thank you.

Secretary Martin. Will it close the— what was your question? Will it close the learning divide or widen it? The digital divide or widen it? It depends how we do it, how we use this.

There is no silver bullet. There is no one tool that is the panacea or the magic way into closing the digital divide. If this is not done well, this is on us as leaders. To make sure that the tools are safe and ethical and used well and the teachers have the proper training. And are they being used across in our state, in Delaware, in rural and every other part of the state?

The tools that have the greatest promise, are they being put in the hands of the teachers so they can use them? But at the end of the day, no tool or curriculum teaches a child. Mm-hmm. The teacher teaches the child. When you equip that teacher with the knowledge and the skill to use these advanced tools, you can actually improve at a great rate the learning outcomes for students, but it's in the hands of the educator, not the tool.

Thank you. This has been an excellent hearing. Thank you. And we went a little bit over time, but I think everybody's interested in this. Wouldn't you like to look into the future of 10 years and see where we're at?

It's going to be interesting. A lot different than when I grew up and a lot of us in here grew up. It's going to be a challenge to not just the student, but also the teachers. To get involved in this. So we want to thank all of you for coming today.

For any senator wishing to ask additional questions for the record, we'll be due at 5:00 PM today, Wednesday, July 1st. You'll probably get some questions on the record. And thank you again to all the witnesses for being here today. We have a letter here I'd like to ask for unanimous consent. It's from the Southern Institute statement for the record.

The rise in AI and education should mean stronger protections for parental rights. I want to thank all of you for being here. Hope we got something out of this. We got it on video. So if you want to change your answers, you— now's the time to do it.

But, but thanks for being here. And thanks for the— all you young people back here being here. This is your future and grab a hold of it and run with it. AI is not coming. It's already here.

Thank you. And learn as much as you can, but use it wisely. So thank you very much. Hearing is adjourned. Thank you.

Good.