Excerpts from panel at the Network for Public Education Conference on the use of Artificial Intelligence in schools, held on Sept. 26 in Houston. The panel led by Leonie Haimson included Mary Burns, a lead author of the Brookings Institute report on AI, Randi Weingarten, President of the American Federation of Teachers, which has just signed an agreement on AI safeguards with Microsoft, and Josh Golin, Executive Director of Fairplay, who organized a letter from over 200 organizations and experts, calling for a two-year moratorium on AI use in schools.
Brookings Institution, A new direction for students in an AI world:, https://www.brookings.edu/articles/a-new-direction-for-students-in-an-ai-world-prosper-prepare-protect/
Brookings Institution one-pager on AI: https://www.brookings.edu/wp-content/uploads/2026/01/A-New-Direction-for-Students-in-an-AI-World-ONE-PAGER.pdf
Randi Weingarten, Devices down, eyes up, hands-on: 10 points to boost teaching and learning in the AI era; https://www.aft.org/press/speeches/vision-public-education
American Federation of Teachers, Devices down, eyes up, hands-on, 10 points to boost student learning and success in the AI era; https://www.aft.org/sites/default/files/media/documents/2026/10-point.pdf
National Academy for AI Instruction, Ai Safety & Privacy Standard For Schools Memorandum Of Agreement, https://www.aft.org/sites/default/files/media/documents/2026/NAfAI-School\_AI\_Privacy\_Standards.pdf
Fairplay, Coalition of Organizations and Experts Letter calling for Pause on Generative AI in PreK-12 schools; https://fairplayforkids.org/wp-content/uploads/2026/05/Fairplay-AI-Moratorium-April-2026.pdf
Fairplay, Statement on AFT/UFT Deal with Microsoft; https://fairplayforkids.org/statement-on-aft-uft-deal-with-microsoft/
Info and sign up for Fairplay's Screen Time Action Network, https://fairplayforkids.org/screen-time-action-network/
Lightly edited transcript of Talk out of School’s Oct. 5, 2026 episode: Expert panel on the what the role of AI in the classroom should be
Leonie Haimson 01:32
Hello, everyone. My name is Leonie Haimson, and welcome to our show, Talk Out of School, on WBAI Radio 99.5 FM and WBAI.org, where we focus on the issues affecting public schools here in New York City, the state level, and nationally. Our show is also available for download as a podcast. Tonight, I'll be playing a slightly edited version of a terrific panel I led last weekend at the Network for Public Education conference in Houston on the use of artificial intelligence in our schools.
The panel members were dynamite, including Mary Burns, the lead author of a Brookings Institute report on AI entitled "New Directions for Students in an AI World. I'll put a link to that report in the resource section of the podcast.
Then we'll hear from Randi Weingarten, the president of the American Federation of Teachers, which has nearly twp million members and more than 3000 local affiliates nationwide.
Finally, we'll also hear from Josh Golan, Executive Director Fairplay, which is the leading organization working to help children thrive in a world that is increasingly dominated by big tech. Each of the panelists spoke for a few minutes about their views on this critical matter of AI, then I followed up with a question or two for each of them. First, let's hear from Mary Burns, lead author of the Brookings Report on AI.
Mary Burns 02:52
I'm a former middle and high school teacher. I've been involved in educational technology for 29 years, and I'm headed next weekend to Turin, Italy, where I'm going to be a visiting professor on AI and education for the University of Turin, 1404, one of the oldest universities in the world. And I'm actually working right now on an EU project examining AI and education policies in 56 countries. I got involved in AI in education back when everybody else did, when it first emerged, and I was in a teacher training hub in Portugal.
I was there for a year, and I actually taught my colleagues how to use AI, and I certainly saw benefits, but mostly what I saw concerned me, and it concerned me so much that I actually wrote a book chapter called "Eyes Wide Open: What We Lose from Generative AI in Education, and the QR code is there. And I mention that because I don't think we're focusing enough on the effect of AI on teachers, and because we were kind of asked to share what our views are on AI. And I think my views have remained fairly static in the last over the last four years. So let me move on again.
So the Brookings report, and I think Randi, you had a hand, Rebecca, and suggesting the title. The Brookings report basically was an 18-month experience. I've given you a handout that really summarizes some of the key points, and I'm going to summarize it to just tell you it's largely a qualitative study. But I did read you know hundreds and hundreds of research studies. We talked to about 505 people. I interviewed about 505 people across 50 different countries. So it's very thorough, and I have to say it's really stood up to a lot of the research that's been produced since. So the report, obviously, you know, we did find some potential benefits and potential risks of AI. Education and I think here the key word is potential because right now the gap between AI implementation and AI evidence base is just enormous and that's normal in technology because it takes years for people to figure out how technology how to really integrate it into assessment systems, into instructional systems, and it takes years to do really high quality research, especially on a suite of technologies that keeps changing all the time. So, I want to point out that here, in terms of these benefits, yes, there are some benefits for English language learners and for students who, especially with the social chat box you've talked about, there are benefits for kids on the autism scale, and I can talk more about those.
But the biggest benefit that we see in terms of AI is really that it's a productivity tool. It's a task replacement tool, so it does, under certain conditions, save teachers time. So we don't have great evidence on benefits and risks, and I'm going to talk about the risks. Although we don't have great benefit, great evidence on the risks, we do know two things are true, and number one, using chatbots leads to cognitive offloading for kids, and that has downstream effects in that it affects children's learning, and secondly, we don't need research to tell us the AI companies are telling us these tools are very unsafe. They pose institutional, systemic, and individual risks to privacy, safety, and security, and we know that from the tech companies, big tech themselves. And this is all made worse by the fact that many education systems across the globe.
I neglected to mention: I actually work across the globe. I've worked in about 100 countries. Most education systems I encountered. We were talking about this don't have AI policies or guardrails in place. So at Brookings, we came up with a framework really around AI use because it's here. It's really hard to put this genie back in the bottle. I think we all know this. So for students to prosper in a world where AI is everywhere and it is everywhere, we have to prepare them and we have to protect them. And when I say we, the report is really a call to collective action from parents, educators, tech companies, and above all, not just state governments. That patchwork of state laws doesn't help. It's really it's troublesome. It's troublesome for technology companies, but we need federal regulations.
So the recommendations are listed. So I won't get into them for you, except to say that one of them certainly focuses on Garfield. Certainly focuses on regulation. We need very strong regulation, like the Europeans have with the GDPR, which we call regulation by design. If you have very strong mandatory national regulation that's enforced, then companies will design their products to those specifications.
Next please. So this is where I break from Brookings because this is now me. So, and it's not to say Brookings doesn't follow this, but I just want to say that I think there's also another really important guardrail that we need to implement, and that's a guardrail around our mindsets. And I don't mean our mindsets. I mean the mindset around AI and education right now. And so right now, what we see is this massive push to innovate, and we know where this leads because innovation that's not thoughtful results in a backlash and overcorrection.
So what I think is, we need the most important guardrail we can have is a mindset that focuses on the duty of care. That our first responsibility as education systems is not to basically ask what can students do, what are students doing with AI, but in fact to say what is AI doing to our students, and above all, we have to make sure that we are not doing harm. So, children have access to these tools. I talked to these kids; they are using them for everything. They are using them for entertainment, for social purposes. They are using them to make behavioral decisions. They are using them to make emotional and moral and intellectual judgments. And as we all know, the use in schools far outpaces the evidence base.
So I think what we have to do in education systems is convince the people who make these decisions and who tell us that we have to go fast to break things. That, as I've said with teachers for the last 29 years, we have to keep two fundamental questions in mind. Number one, does this help my students learn something that would be otherwise impossible. And number two, does it help them learn it better? So, if we're really going to go to the heart of the Brookings report, I think it's this is a message that the future of AI is not going to be shaped so much by the technology. It's going to be shaped by the collective choices that we all make about design and governance and use and accountability. So with that, I don't think you got the original. This is a QR code. If you feel like reading the actual report, I'm going to add an AI summary for you. And otherwise, thank.
Leonie Haimson 11:05
Next, let's hear from Randi Weingarten, the president of the American Federation of Teachers.
Randi Weingarten 11:10
So, Rebecca and Mary are my teachers on this, and you know Rebecca is Rebecca Winthrop from Brookings, and I have had many a conversation about this. Let me just say I want to I want a plus plus, and I'm sorry I missed some of Leonie's opening. I want to plus plus what Mary said and fair play, and we have worked on a lot of child safety work as well, and I'm a plus plus a lot of what Josh says. We got into this work on AI because of several things.
Number one, I did not want. I wanted to do whatever we could to not have the mistakes of social media repeated. But the reason we gave the speech I gave last May that was called in, and I'm sorry, I didn't bring stuff with me. That's or did we bring stuff in there? Remember, it's called devices down, eyes on, hands on. It's like 10 points, and if you look at L.A. and you look at New York City, a lot of their policies look like what we proposed at the end of May. But that was the work that Mary did, and so Brookings did that, Jared Horveth did, that others did on education and on cognitive offloading completely scared me, and I thought that if we don't actually say something as a union, regardless of what gets visited on us, because edtech has really like Zeph [Capo, President of the Texas AFT] talked a lot about privatization today and the effects of the starving, the sneering, the shaming, and the shuddering.
I think I added another S. But what has happened? Not just because of COVID. COVID accelerated it. But what has happened is that ed tech became the panacea for schooling, and there is millions, billions, and billions of dollars that have been in all of our schools, frankly, public and private, based upon edtech, and it became the you know the the silver bullet, and what has happened and Leonie said this earlier the amount of AI that is now incorporated into these tools no one knows parents don't know teachers don't know.
I mean would it surprise you that this summer Google said that Gemini is now in every one of its Chromebooks, in every one of its screens, and no one knows what the level of which they can actually take children's information and use it and use it in any other kind of way. So this but what really scared them was the offloading, and that okay, shame on a college student for letting AI offload their education. They can make their own decisions, but for a kid who. Three or four or five or six or seven or eight-that's our collective responsibility. And if we don't actually utilize that, that and we don't actually understand that kids are basically not really engaging in critical thinking or relationship building because of this. I'm not even talking about well-being. I'm talking about basic acquisition of knowledge and application of knowledge, and that's what actually moved me.
So thank you, Mary, and thank you, Jared, who I said thank you to yesterday in the panel, because we don't have enough research, which is one of the reasons why I call for research and an AI research consortium. So that's what kind of motivated us to do what we did, both with the AI Institute with the devices now speech, but also in that speech, I said to the tech companies, I said very publicly to the world, including the tech companies we were working with, we will not work with Volunteer. We will not work with Musk. I've said to the companies that we were working with, and you know, I'm not sure that we should be working with OpenAI, but we are.
But I said to all of them, if we don't get to a safety standard, we will shut the doors of our AI institute. As important as it is in terms of teaching teachers, so that we empower them to do their own advocacy here, we will walk away. And given Gates understood, Microsoft understood that I was serious about it because we have walked away from millions of dollars of a Gates grant because of not only of what they did with info, all sorts of other things.
So they knew that I was serious, and all summer long we spent, and I spent I spent my summer negotiating with Microsoft to basically get to what is now our first in the nation safety standard. Now I ensure it's not perfect, and I would much rather have a safety standard, a data safety standard, in law in FERPA than as procurement policy with a company. But in the absence of having it in law, I didn't. We wanted to make sure that there was a way that we could leverage this institute, this little institute that we have to train teachers to actually create procurement policy in the nation.
And what has happened is everything they have agreed to do with us is now going to be procurement policy for every single school that they have contracts with starting November 1, and that policy is 30 pages. I'm sure somebody can find something that is wrong with it, but it's 10 principles, 30 pages. A lot of it based upon the learnings that we have from Leonie, and that we have from Fairplay, and that we have from Brookings, and from so many others, since people really advised us, including we have one of the most successful lawyers for child privacy who's been working with us side by side.
So the bottom line for us is, and that we have an obligation to find a way to navigate this to navigate AI not only in schools but in society, and for us safety is more important than the innovation for us, safety for children-not just well-being, but safety in terms of every aspect of education-is really, really important. And the second piece that's really, really important is safety….Let's let me move from education to issues about cyber security, security, worker displacement, climate. All of those issues are really really important. The one issue I do think that we will make real progress on, and a lot of that is because of the work that all the three other colleagues on the stage are doing.
The one issue I do think we will make progress on in the next Congress is children's safety, and is and I think we have to do whatever we can to force the federal government not to do this excuse my language preemption bullshit, but to actually make sure we can get these guardrails into law as fast and as effectively as possible. I don't know what AI looks like in five years or in 10 years from now, but and I was saying this to Leonie earlier. I learn something new every day. But what I do think we need to do is we need to put as much information in teacher and parents' hands as possible, and the more information we can put into teacher and parents' hands, the more we can not only protect safety, privacy, and security of children, the more we can protect their well-being, like having understanding the sycophancy of these chatbots and trying to actually prohibit their you know use for kids under 16, which is what New York State law is. I mean, it's on it's on Kathy Hochul's desk. It's something that Common Sense Media and others have really pushed for. Sure, you push for it too.
The laws that Fairplay have gotten done in California, in particular, are some of the best laws in the nation. But we need to actually deal with privacy, well-being, and knowledge all at the same time, and so we are also pushing very hard not only for an analog system of knowledge, particularly in terms of pre-K through either elementary or eighth grade. I say analog because we're pushing for doing some of things like critical thinking in the old-fashioned way, like doing project-based instruction, doing experiential learning, finding ways to do all that.
But we're kind of going back to basics of what should schools be. We need classrooms to be safe and welcoming for children, and we need curriculum to be engaging, relevant, and rigorous, but engaging starting first. Relevant is something that kids will resonate to, and that's why I called my speech devices down, hand eyes up, hands on. So there's a whole bunch of other things in there, but I just really wanted to make sure you saw the theory of what AFT is doing.
We have a multi-word, multi-page resolution that we passed at our convention this summer on everything related to AI. And I think finally, this is the last thing I'll say right now. Finally, I think the Anthropic researcher leaving, quitting Anthropic, and saying that these that we have they have lost control of the tool, so that the tool actually is generating itself. I think that finally got into the cultural ether. So for the first time, you saw what just happened in Florida. We've had the same similar conversation with the governor in in Massachusetts. I see Julian back there. If Jocelyn Benson wins as governor in Michigan, she is very focused on doing something very similar.
I think what we're going to see is governors wanting to actually regulate not only on the safety issues when it comes to children, but also on the education issues and also many of the other things, not just moratorium, but regulate on this. I think Pramila Jayapal's idea about chartering AI companies is a really good idea. I think the tech tax ideas that are coming up are really good ideas. But this is a very fluid situation in terms of AI, and I think the more we know, the more we will be able to actually advocate. And so I am advocating for engagement, not for trying to, not for being so fear-based that we just say let's have a moratorium forever. That's but we really have to create the safety guardrails that on all of these different issues, including on education. The most asked for course in our AI institute is the one that says cognitive offloading and critical thinking in an age of AI. Teachers really, really, really, really want to figure out how to offset this cognitive offloading that is happening with kids in every other environment where they're using AI, and how to actually really engage in terms of.
Leonie Haimson 24:58
Finally, here's Josh Golan, Executive Director of Fairplay.
Josh Golin 25:03
Thanks so much. It's great to be here on a stage with three people whose work I admire so deeply. I'm going to start by Fairplay is a national advocacy organization, and really our guiding principle is that the people and institutions that influence children should be people and institutions who have children's best interests at heart, and that sounds pretty simple. But we all know that children are incredibly influenced by social media, by what they're doing online, by these online platforms, who don't set out to hurt children, but are pretty much neutral on whether things are going to be good for children or not, and their number one mandate is to make money off of children.
And we've seen time and time again in the social media context, thanks to whistleblowers and thanks to documents that we have from lawsuits that when it comes down to choosing between safety and profits at these social media companies, they choose profits every single time, and the reason they're able to do that is because we're terrible when it comes to passing laws to protect children. The only federal law that we have that protects kids online is the Children's Online Privacy and Protection Act, which was passed in 1998. Before YouTube, Facebook, smartphones, AI were even things that anybody was were talking about and didn't exist in real forms. So we know that introducing children to the most powerful persuasive technologies without adequate protections through policy is a disaster. We've seen that in social media and teen mental health crisis, which has been driven by social media. Not to mention all of the kids who have died because of things that happened on social media. I have the great privilege of working with families who advocate for social media protections because their kids are no longer with us.
I also think it's incredibly important that when we talk about AI in schools, that we do it in the context, the greater context of ed tech in schools, and it is so important to do that because, first of all, we're talking about a lot of the same players: Google, Microsoft. You know, those are those were very much at the forefront of the move to transform education to being screen based and one to one devices, and what we now see in schools. The same promises that that you know one-to-one devices and ed tech. We're going to revolutionize education. We're going to personalize education somehow by using technology. We're going to make education more personal, and that it's going to free up teachers to do to be to focus on the things that they really want to focus on, get rid of their mundane tasks. That is going to reduce educational inequality between lower-income school districts and kids from wealthier. None of that has come true. The achievement gap is as great as ever, and in fact, what we see is that schools and school districts and countries that introduce educational technology much more slowly and thoughtfully, rather than just designing the curriculum around a device, that kids have much higher test scores in those countries, and so the promise of ed tech was very much a marketing gimmick, and so much of what they promise does not come true, and so I think it behooves us to think about those very same promises that are being made often by the very same companies with AI.
We know from ed tech that asking kids to learn on the very same devices that they use nonstop for entertainment doesn't work. That if you give a kid a laptop where they watch YouTube videos at home, they're probably going to be doing that same thing while they're in class, even if the school blocks it. They kids are pretty good at getting around those blocks. And in fact, you know, we know increasingly from people like Dryer Horath that you know, and research that kids are off task in school when they are on a device almost 50% of the time, and I don't blame them.
I mean, if I had been 13 and in school and had access to have private sneaky conversations with everybody I knew, to play every video game in the world, to watch every movie that was ever created on my lap, I probably would have done the same thing, and I don't envy teachers trying to teach that. And I'm so glad that we now have bell to bell policies in almost 30 states in this country, or laws that require bell to bell policies in almost 30 states in this country, so that we're taking the success. Phones away from kids. That's enormously important.
And then I think the last thing, the sort of parallel that I want to talk about with ed tech and AI, is this sense of urgency. This idea that this is such a powerful and perfect tool that if we don't start incorporating and redesigning our curriculum entirely around AI. We are literally hurting students, and that's quite a statement to make. And I want to just unpack a couple of like pieces of that. Like one is like we're going to lose out to China if we don't do that. And I'm sorry, but I just don't understand how if a second grader is using Magic School or not? That's going to affect things like global competitiveness and the race for economic dominance.
The other thing that we hear is that we need kids using AI in schools to prepare them for the jobs of tomorrow. Well, that's a wild claim because four years ago nobody was saying that we should teach AI in schools to prepare kids for the jobs of tomorrow. So I don't think in four years we particularly know what we're going to be claiming the jobs of tomorrow are, or in 10 years, or in 15 years. So this and you know it's funny because when I give when I give talks, there is inevitably a guy in the back of the room who raises his hand-it's always a guy-and he says, you know, I'm printing money using AI. I have, you know, I'm building apps every day using AI, and if we don't have kids learning this stuff, we're going to be so far behind. And I look at him and I'm saying, you know, you look like you're 35, 40. Did you learn AI in school? He's like, well, no. But so, so, so the idea that we should be teaching this at a very young age, instead of things like critical thinking, instead of learning hands-on learning, it is a it is a marketing device. It is it is driven by the AI companies.
And so, what I really want to say about this urgency thing, because I think this is so important, the question before us, from my perspective, is not does AI have any benefits for learning? Will AI ever have any benefits for learning? The question before us is: Is the AI that we have today, with the teachers and the training that they have today, with kids and their propensity to go off task, and the fact that they're already using this stuff in ways that are really concerning, like confiding in AI chatbots, like asking for advice about their most sensitive information and their problems, is that AI ready for prime time, and I would say you know based on the things that that all my three colleagues up here have said today, and I'm not going to go through them necessarily all again because I think we've talked.
I want to get to the discussion, but I think it's very clear that when we have this cognitive offloading problem, that it's not ready for prime time. When we have this enormous privacy problem where we don't understand what's happening with the data, the reading. There was a great piece yesterday in one of the San Francisco papers about the reading program Amira and how the privacy policy doesn't even say that they keep voice recordings of kids, even though that is central to what they're doing.
So, so I think when we think about the privacy problems, when we think about the enormous data security problems, when we think about the fact that this is really a further move, from my perspective, to deprofessionalize the teaching profession, to make teachers much more about being data collectors and data analyzers rather than actual pedagogues, I think that there is very clear evidence that this stuff is not ready for prime time, and so what Fair Play did is we put together a call for a five-year pause on the use of generative AI in schools.
Leonie asked, "How did we arrive at five years? Well, we made it up. I'm sorry, but you know, part of the idea was at the time that we were writing this, the AI companies were trying to get a 10-year moratorium on states being able to regulate AI. So basically, no rules for 10 years from the states, and we said, well, if they're going to be greedy, like we're being half as greedy as the AI companies, we're only asking for five years. So, so we put out this call for a five year moratorium, and then we and was worked on. So we have at Fairplay we have an Action Network with different work groups around different topics, and we have a screens and schools work group. And so several members of that work group worked on our AI moratorium statement, or AI pause, is what we're calling it. And then we shopped it around to organizations and to education experts and asked them to endorse it. And we got more than 200 leading organizations and experts to sign on. And we see this really as an organizing tool.
We were thrilled to see it referenced in New York when City Councilors were supporting the AI moratorium in New York City. We really see, like, for people to be able to go to their school districts and ask for a pause on AI and say, "This is not just me making this up on my own. This is this is something that this is policy that is endorsed by leading experts all over the country, and we're thrilled to see that that parents are using it in that way. The last thing that I'll say is we are at a remarkable political moment on AI.
Leonie showed some polling before. There was another poll that came out just a couple weeks ago, that by a margin of three to one, parents support a moratorium on the use of generative AI in schools. That is a remarkable number, three to one, and so I think what is happening right now is a real test for our democracy because what is what is really going on is that these companies have so much investment and they need to turn a profit. And one of the ways, and we know this, we know this from the ed tech contests too, right? So one of the reasons, and Natasha Singer at the New York Times has done such fantastic reporting on this, but the drive to one-to-one devices in schools was not driven by educators thinking that if kids were on screens all day wearing headphones and not talking to the kid next to them or interacting with their teacher, that that was going to improve education. It was driven by the fact that Apple, that Apple and Google are trying to get lifetime brand loyalty to either Apple or Google products, and one of the ways that you get lifetime brand loyalty is you put things in schools and you require people to use it, so then they become used to that system.
The same thing is absolutely going on with AI right now. You know, there is trillions of dollars in investment in these companies, and they need to recoup it. And one of the ways that they are going to recoup it is by having kids have lifetime loyalty to Claude or ChatGT or Muse or whichever AI system that ends up. You know, they all they want to be the what Google is to search. All of these companies want to be the chatbots, and one of the ways that they do that is by getting into schools and creating that brand loyalty.
So, so ultimately, I think when three-to-one parents want to pause, this is really a test for democracy about whether these incredibly powerful companies, backed by all the venture capital in the world, get their vision for schools, or whether and whether we want corporations deciding what our kids are learning in schools and how they're learning, or whether we want educators making those decisions and communities making those decisions, and again, this is not a question of whether AI has any good purposes or whether or what AI will look like in 10 years. It is very possible that we will learn how to make AI in ways that it doesn't form unhealthy attachments between kids and the technology, but only if we have policy. Like that's not going to happen on its own. But maybe we will get those policies that can make this stuff a lot safer and reduce the negative the negative consequences of AI and really help kids learn, but we are absolutely not there yet. And so, to act like we have to do this, or we will be harming schools, is a travesty.
And so, again, I think you know, based on all of these concerns, the there is a clear case to be made for slowing down, training teachers on how to use this stuff, understanding what are the really good uses of it and what are the most harmful, getting clear policy in place so that data is protected, so that parents have rights, so that we understand, so that the research can catch up and we can understand what works and what doesn't work, but to remake education in order to recoup trillions of dollars of venture capital has been invested in these companies would be one of the greatest travesties in the history of education from my standpoint. So I look forward to the discussion.
Leonie Haimson 39:17
After the three panelists spoke, I followed up by asking each of them a couple of questions.
So first, Mary, given all the troubling findings in the report, why did the report's recommendations fall in this order-Prosper, prepare, and protect- rather than Protect, prepare, and prosper? That's my first question.
My second question is: All these states and districts are putting together their policies now and their guidance. Have any of them reached out to you and other researchers in the field who've done so much work to ask you for help in preparing this guidance?
Mary Burns 39:56
Thank you. So I can answer the second question briefly. To be, I don't know because I had a short-term contract with Brookings, so I'm not sure what you know is going on there. Yeah, and the first question: Prosper prepare and protect. That was a lot of internal discussion at Brookings. I actually thought protect should go first, also as well.
But I think the logic is: look, if these tools are here, we do want students to use them and prosper to thrive. You know, thrive emotionally, thrive cognitively, certainly thrive vocationally. And so we have to prepare them. And I agree, a lot of preparation has to be not necessarily technology itself, but really the domain knowledge and critical thinking, which goes back to domain knowledge, because you have to be critical, a critical thinker about something. But then there's so many protections as well.
So, and I think probably some of it. This is getting under the hood a little bit. Is we had a task force, quote unquote. So everything Randi probably knows better than I do how virtual works, but my understanding is when they come up with something big as a task force. So I said, well, everybody we talk to is part of the task force, right? The mother in India, she's going to be on our task force. Our task force, though, I think sometimes gets narrowed down to we had a steering committee, and it was made comprised of 40 people, who I presented everything back to these folks, and a lot of them were AI and education people, and you know AI thinkers, and so they're going to have a different perspective. So I think some of that is that kind of internal negotiation that goes on.
Leonie Haimson 41:45
Yeah, well, thank you for that.
Randi. So I have two questions for you as well. First, you know, a lot of people, students, parents, and teachers themselves are still concerned about the use of AI by teachers themselves for lesson planning, feedback to students on their work, and worry that this might erode the connection between teachers and students, and even breed resentment among students and distrust, since they don't really know who's giving them feedback and who's communicating with them. Is it a teacher or is it a bot? And so none of the states and districts that talked about it put really any real guardrails on that, except for the thing about grading and high-stakes decisions.
So I'd like you to talk a little bit about that first. Why you think that there should not be stronger restrictions on teacher use of AI, and then I have a question about this Microsoft agreement you have. I must admit I haven't read it carefully, but I did look over it and I didn’t see strong enforcement and oversight in it.
And I'm very concerned because all these companies, all these AI companies, have been known to lie repeatedly about the issue of data mining, and most recently, there's a court case in California that was filed against Microsoft and others. And despite their claims that they do not plagiarize or data mine copyrighted material, documents that were gotten through the court case through discovery show that a Microsoft director called what they've done "quote the largest theft of labor in human history,” and another executive at Microsoft, had privately said “Millions of people around the world will soon consider large models hoovering up all their work to be an astonishing theft of unprecedented proportions.” So why did you give so much trust to Microsoft in this agreement, and do you think that we maybe need more oversight and real guardrails and real enforcement mechanisms as well? Because I think it's difficult to know what they are doing. We don't trust these. I mean, I don't trust them. I think a lot of other people don't trust these companies and they're worried.
Randi Weingarten
So, page 10 of this 30-page document has amazing number of third-party audits that are part of this agreement, and one of the things, since I can barely use my own device, I
one of the things that we did was we actually used an outside lawyer, Mark Mayo, who has done many, many, many of the cases against these companies. Who is our lawyer here in terms of intellectual property? And I wanted to pick somebody who was very adversarial to these companies because I wanted to pick somebody who was like the best and. You had to see a bunch of these lawyer conversations. I mean, I don't. I try not to practice law very much anymore.
But literally all summer long, I was involved in these conversations, and the key was in terms of Microsoft because I don't trust these companies either. But the key. This is what the. This is what we ended up really thinking about over and over again. Josh is right that we've been fighting their attempt to get a 10-year moratorium on any state doing any rules affecting AI, and there have been three or four times over the course of the last year or two, last year, that they tried to put this moratorium, their tenure moratorium, as part of like must pass legislation, like the Defense Department legislation, like the funding the budget, and you would see people like Fairplay and us basically saying, "Well, no. And sometimes Steve Bannon would actually help us, and others the MAGA movement, the MAHA movement, would actually help us. And there's a bunch of people who say, "How can you stand with people from Moms for Liberty or MAHA?
I'm like, if they're going to help us get this stop this attempt on preemption, I'm going to stand with them. We have to stop this because that's the big tech. So the what I think all three of us see because we spend a lot of enough time in Washington is that big tech, particularly Musk and Palantir. Just Google this name, Curtis Yarvin. Just Google the name. They have a theory of the case here that I would say is more than brand loyalty. They have a theory of the case here that is available, just like Flock cameras, that they will use AI to surveil the country and that they will actually replace democracy with AI and with surveillance. They do not believe in democracy.
So the thing that I am very focused on is we are in a race with this. It's fantastic that public opinion has changed, but we are in a race, and so I wanted to figure out a way that this little AI institute that we have, all these companies were very happy to give us a privacy agreement on what we did in the institute. They would have been very happy to do that. That's not what we bargained for. What we bargained for as a third party, was how do we create procurement policy so that this is your policy with school districts that that you have to give parents information about what you're doing in a school district that you have to say in that school district that it is your policy not to data mine, not to use kids as Beta testing, not to do all of these things.
Everything we could figure out what to say is to try to protect privacy, to try to update FERPA, to try to look at what Mary and Josh and Jared and so many people. We tried to do this as part of the 10 principles, and what took the most amount of time was that we they would have given it to us in a minute for the AI Institute. We said no. This has to apply to school districts. So in this contract agreement with Microsoft, it says if the school district asks for it, they must provide it. But then what they have now done is they have now said we will make it part of our contract November 1, and that's because of the public opinion. We could contract with them to say, and then we would be working with all of you to say no. You need to do this, and we're creating a piece of paper that says what Microsoft can turn off and what school districts can turn off. Because you know, school districts can turn off a lot of the shit. They can turn off a lot of the things that the that this does.
But none of us really know because school districts look at Microsoft and say, "Oh, it's theirs, and Microsoft looks and says, "Oh, it's yours. And Google is even worse. So that's what we try to do. Is it perfect? No. Does it need to be in law? Yes. But in the absence of anything else, we wanted to start doing this. We wanted to start, and we thought that you had to either do it with a Microsoft or a Google because they are the ones who are so prevalent in schools. And the leverage we had was they did not want at this moment politically.
I think they did not want me to say we're walking away from you, and I think that was the leverage to do. Just to be honest, the leverage we had now in terms of teachers and AI, what you see actually here is an ethical. We have had lots of policy within our own union about the ethical uses for teachers and AI, and so part of it is basically AI cannot be the decider; the teacher has to be the decider. So it's not just on grades; it's on even grading papers and all that. So the bottom line, and that there has to be an FO use, and part of what we're doing in the AI Institute is really, really drilling that down about the FO uses and what your role as a teacher is, and that is actually part of what we teach everyone who goes into the AI Institute in terms of that. And the AI Institute is really oversubscribed.
Leonie Haimson 51:20
So just to just to clarify, students' papers with the comments and corrections and all that kind of feedback will still be done with AI, but with teacher oversight, or supposed to be done by the teacher herself?
Randi Weingarten 51:22
What I'm saying is, the teacher has to be the decider. Now, I there are there are people who use personally. I just taught a college class up at Cornell a few weeks ago. I grade my own papers. I just think it's important. I know people who do a first pass using AI. We have policy as we are not the employer of teachers. We believe teachers, not AI, must be the decider on all of these issues. Will people personally? I don't think people should be using AI personally. I don't think people should be using AI to write papers. That's me personally. But we have said ethically, teachers must be put to the side. So now a question for Josh, and we'll open it up to the audience. Parents and teachers are incredibly busy, and so many different responsibilities. second. Let me just say it's on page. For those of you who have it, it's on page nine in terms of what the academy does in terms of all of this stuff.
Leonie Haimson 52:53
Josh: How do parents, teachers, and others who want to get involved in this issue focus their efforts at this point in time, on their districts, on their state levels, on the national level, through legislation, through regulation, through district policies? Where can people's voices be most effective, and where do you see the hope in the future to get real guardrails and restrictions on the use of AI in the next one to three years?
Josh Golin 53:26
That's a great question. I mean, I think given sort of the urgency that I talked about and how quickly this stuff is moving, unfortunately, there's no time for federal legislation that takes that will probably take far longer than we have. I think the most I think what we've seen in New York, in L.A. what we're seeing in Chicago, what we're seeing in a lot of smaller school districts is that the district level is probably where the most impact can be had at this moment. I'm also hopeful that we're you know in New York, we just had the Focus Act introduced, which would severely limit the use of AI in schools. And so, I think state legislation we may have opportunities in in in some states as well.
For people who want to get involved, FairPlay has an Action Network. You can Google it. We have a screens in schools work group, which is really active, and it's a place where people who are working on local issues come together and share strategies about what's working and what's not working. There's also so many great groups now that are doing work to try and change policy in their own districts or at the state level. Distraction-free schools being one of them. Schools beyond screens. I'm not going to list more because there's so many.
But I think one of the things that's happened in the past few years is that this has become a real issue of parent organizing, and in a way that we couldn't have dreamed of. You know, when I was doing this work even five years ago, and so what you are seeing is parents and teachers and public health professionals coming together and organizing in all sorts of really creative ways and demanding these policies. I think they're taking school boards a little bit by surprise, sort of with the vehemence in which they're this is such an important issue to them, but again, I think you know the fact that we're not asking for no AI in schools ever, but we're asking to put the brakes on a on something that is going way too fast.
Leonie Haimson 55:37
This is Leonie Haimson, host of Talk Out of School on WBAI 99.5 FM Pacifica Radio. I hope you enjoyed our show, which is also available as a podcast. If you missed the live version or want to recommend it to a friend, if you go to Simplecast, you can find more news and information on the issues that we covered. Also, please consider becoming a member of WBAI or a special supporter of Talk Out of School by donating at wbai.org. There's no other show on the air that delves into the many issues and controversies affecting our public schools like this one. We'll be back next week with another episode of Talk Out of School, and thanks so much for listening.