How AI Is Replacing Children's Ability to Think (with Randi Weingarten)

Summary

---

Future of Life Institute Podcast

Episode: How AI Is Replacing Children's Ability to Think (with Randi Weingarten)

Date: June 30, 2026
Host: Gus Ducker
Guest: Randi Weingarten (President, American Federation of Teachers)

---

Brief Overview

This episode tackles the growing influence of artificial intelligence (AI) in education and the troubling ways it's starting to undermine children’s critical thinking, attention, and learning capacity. Guest Randi Weingarten, a leading voice in American education, outlines both the practical dangers and philosophical dilemmas created by AI use in schools. Drawing from research, policy experience, and on-the-ground feedback from teachers and parents, Weingarten calls for a robust set of guidelines, privacy protections, and a cultural shift back toward active, experiential learning. She delivers a passionate call to action for policymakers, educators, and parents, warning that children are at risk of losing the "muscle" of thinking and learning if AI and screens are not properly reigned in.

---

Key Discussion Points & Insights

The Current State: "Wild West" of AI in Schools

• No clear regulations or state/national guidance.
  • Kids are a "prompt away" from knowledge, making it hard to assess if they've truly learned (00:00–03:00).
• Prevalence of screens and AI has correlated with declining test scores and diminished attention spans.
  • Research (Jonathan Haidt et al.) shows cognitive, attentional, and persistence challenges increasing as screens proliferate.
    • "The test scores in America and the world have gone down from the time that screens have become really, really prevalent and accessible for kids…if that is the case, that is a five alarm fire." – Randi Weingarten [05:35]
• Absence of leadership:
  • No significant action from the US President or Secretary of Education; the environment is described as "the wild wild west" and "Lord of the Flies" (00:00–01:40).

---

The Harm: How AI Can Undermine Learning

• Cognitive Offloading:
  • AI allows students to outsource thinking—using AI to write essays or solve homework leads to less actual learning (12:18–14:42).
    • "There is now a new term for this, which is called cognitive offloading…a student uses a prompt and AI to gain information as opposed to figuring it out themselves." – Randi Weingarten [12:18]
• Unsupervised and poorly understood deployment:
  • Many teachers and parents are unaware if and how AI is being integrated into students’ devices—e.g., Chromebooks with AI-enabled features rolling out without consent (09:58–12:07).
• AI can help experienced adults but harms novices:
  • For adults already versed in a topic, AI can augment learning or offer new perspectives. For novices, it's a shortcut that bypasses genuine comprehension (13:00–14:42).

---

The Solution: Guardrails and a Return to Active Learning

• 10-Point Plan for AI in Education
  • Developed with reference to research and stakeholder feedback, advocating for stronger boundaries and focus on deep, meaningful learning (04:30–07:00).
• Definition and Importance of Active Learning:
  • Emphasis on "project-based, experiential, learning by doing" methods (07:27–09:43).
    • Examples: Debates, hands-on career tech education, music composition, and collaborative projects.
    • "What you're doing is you're helping kids learn not only what the information is, but how to apply it, how to use it, how to work with each other." – Randi Weingarten [08:34]
• Ban screens K–2 except for compelling reasons:
  • Young children's brains are especially vulnerable, and early habits of real-world attention and persistence should be prioritized (22:13–24:00).
• Limit child-facing AI in elementary education; no social chatbots under 16:
  • Safeguarding children’s development, social skills, and vulnerability to manipulation (26:06–28:25).
    • "The chatbot becomes your companion... when it keeps telling you 'I love you', it keeps telling you you're fine, it keeps being sycophantic in a way that you then are addicted to it... It doesn't give you any resiliency or any sense..." – Randi Weingarten [26:22]

---

Institutional and Policy-Level Solutions

• Call for a nationally- or philanthropically-funded, independent research consortium:

  • To systematically research both harms and best practices, as opposed to relying on fragmented or industry-funded studies (20:07–22:13).
  • "We need good research... so that's part of the reason why I think in schooling we have to make schooling interesting, but we have to be able to teach kids how to learn." – Randi Weingarten [21:18]

• Gold standards for AI privacy and safety:

  • Parental control over children's data, transparency in collection and use, and strict privacy protocols (30:11–32:33).
  • "We need a standard to ensure to parents that your kid's data is protected. We need to make sure the teacher's data is protected… the parent didn't give that permission." – Randi Weingarten [30:20]

• Industry relationships circumscribed by ethics:

  • Collaboration with Microsoft, OpenAI, Anthropic, but refusal to work with companies they believe fail in child protection (41:14–42:56).

---

Old Meets New: Analog vs. Digital

• It’s not about eliminating technology, but about balance:
  • Advocates some "old-fashioned things"—paper and pencil, blue book exams—while embracing technology's appropriate role (15:09–17:05).
  • "Technology is not going away... But we have to figure out ways to do some of the old-fashioned things like teaching with paper and pencil..." – Randi Weingarten [15:09]
• Teaching digital literacy and AI ethics as children mature
  • Analogized to driver’s education: no one drives without training, so why let children use powerful AI unsupervised? (23:41–25:55)

---

Societal and Psychological Implications

• Risks of AI-simulated relationships:
  • Children may form unhealthy, even addictive bonds with chatbots, at the cost of learning real-world resilience, emotional skills, and critical thinking (26:06–29:23).
  • "It simulates hearing what you want to hear... it doesn't give you any resiliency or any sense. And it makes a premium. It says that the isolation is okay." – Randi Weingarten [28:28]
• Potential for manipulation and propaganda:
  • If a chatbot gains a child’s trust, it could be used to manipulate or even cause harm (29:09–30:11).

---

The Road Ahead: Vision for the Future

• If these reforms are implemented:
  • More vibrant, engaging, responsible education where students have agency, deeper understanding, and teachers experience more satisfaction (43:03–43:55).
  • "You'll see a lot more agency for children, that they will feel more responsible for their education. They'll feel more excited about it, more engaged in it, and more confident about it. And I think teachers would like it a whole lot more." – Randi Weingarten [43:22]
• Ongoing need for shared learning and adaptation:
  • Encourages regulatory elasticity to allow for context-specific decisions, but within established guardrails (36:10–41:14).
• Ultimate message:
  • To truly prepare children for an AI world, we must teach them higher-order skills by sometimes intentionally limiting exposure to the very technologies they’ll interact with as adults.

---

Notable Quotes & Memorable Moments

• "The test scores in America and the world have gone down from the time that screens have become really, really prevalent and accessible for kids…if that is the case, that is a five alarm fire."
  – Randi Weingarten [05:35]

• "There is now a new term for this, which is called cognitive offloading… a student uses a prompt and AI to gain information as opposed to figuring it out themselves."
  – Randi Weingarten [12:18]

• "What you're doing is you're helping kids learn not only what the information is, but how to apply it, how to use it, how to work with each other."
  – Randi Weingarten [08:34]

• "The chatbot becomes your companion... when it keeps telling you 'I love you', it keeps telling you you're fine, it keeps being sycophantic in a way that you then are addicted to it... It doesn't give you any resiliency or any sense..."
  – Randi Weingarten [26:22]

• "We need a standard to ensure to parents that your kid's data is protected. We need to make sure the teacher's data is protected… the parent didn't give that permission."
  – Randi Weingarten [30:20]

• "You'll see a lot more agency for children, that they will feel more responsible for their education. They'll feel more excited about it, more engaged in it and more confident about it. And I think teachers would like it a whole lot more."
  – Randi Weingarten [43:22]

---

Important Timestamps

• 00:00–03:00 — Overview of regulatory (in)action and AI as the “wild west”
• 05:35 — Screens’ correlation to falling test scores: “a five alarm fire”
• 07:27–09:43 — Definition and examples of active learning
• 12:18 — “Cognitive offloading” and how AI enables shortcuts over learning
• 22:13–24:00 — Rationale for banning screens in K–2
• 26:06–29:23 — Dangers of AI social chatbots and simulated relationships
• 30:11–32:33 — Call for gold standards in AI privacy and safety
• 41:14–42:56 — AI company attitudes and ethics
• 43:03–43:55 — Vision for a vibrant, agency-driven future classroom

---

Summary

This episode delivers a nuanced, urgent appraisal of AI's double-edged potential in education. Randi Weingarten argues that current trends threaten children’s mental muscles of learning and thinking, requiring strong limits on early exposure, privacy safeguards, and a systemic return to authentic, active learning. The conversation balances practical recommendations—like K–2 screen bans, limits on AI chatbots, and privacy standards—with philosophical reflections about childhood development and societal responsibility. Rather than advocating for a retreat from technology, Weingarten insists on a thoughtful equilibrium—leveraging technology for teachers and mature students, while rigorously safeguarding childhood learning and wellbeing.

For those seeking a roadmap to navigate AI’s impact on the next generation, this episode provides both a broad strategic lens and concrete policy ideas—along with a call to action for collective responsibility across institutions, policymakers, and families.

Transcript

[A]

Here we are dealing with AI without actually having the balance right in terms of what AI does, what it means to our kids, what it means to the economy. And we don't actually have in the United States a president or a Secretary of education who has any interest in trying to help navigate through all of these issues. So it's basically the wild wild West. If you are a prompt away from just lots of knowledge, then how do you actually assess whether kids have learned anything ban except for compelling reasons screens K through 2. Stop with actually having child facing AI or AI in children's education in elementary schools. Frankly, this is the irony. In an AI world, kids really need to be prepared with critical thinking. They really need to be prepared with understanding context, with understanding what's going on around them. They really need to be able to think and problem solve and do higher order skills.

[B]

Welcome to the Future of Life Institute podcast. My name is Gus Ducker and I'm here with Randy Weingarten. Randy is the president of the American Federation of Teachers. Randy, welcome to the show.

[A]

Thank you. It's really my honor to be here.

[B]

Great. You recently gave a speech at the National Press Club talking about AI in education. Maybe let's start there. What was that speech about?

[A]

So, you know, we are facing seismic changes not only in technology, but the way in which technology is operating in the country and in the world. And we never quite got right the balance between the harms and the positives of social media. And so here we are dealing with AI without actually having the balance right in terms of what social media does, what AI does, what it means to our kids, what it means to the economy. And we don't actually have in the United States a president or a Secretary of education who has any interest in trying to help navigate through all of these issues. So it's basically like a Lord of the Flies situation where, you know, whoever, it's the wild wild west, it's whatever, whichever, whichever metaphor one wants to do. But it's kind of sort of this disruption in, is omnipresent. What we thought we needed to do is to think about the question of what do students need to be able to do? What are the skills and knowledge they need in this moment when they graduate with this huge disruption of technology? And what kind of guardrails need to be there in terms of education to protect people, their privacy? So what do we need to do academically and what do we need to do to protect them? And in the absence of any literally state or national guidance on this, we really started diving into what people were saying and doing, including from as early as last August, talking to Jonathan Haidt about his new research and what it meant. And so what we produced in the end of May was literally a 10 point plan to say this is the broad guidelines of what we should be doing in teaching and learning in this AI age. The linchpin for me was this Jonathan Haidt and Jared Corvette, cloney and so many other researchers have started understanding, and the research seems to support it, that the test scores in America and the world have gone down from the time that screens have become really, really prevalent and accessible for kids. And so this, there seems to be this limit on attention, on cognition and on persistence, as we can see from test scores. If that is the case, that is a five alarm fire. It means that we in education need to do things to help ensure that the muscle of attention, the muscle of cognition, the muscle of persistence, not just well being, as important as well being is, is attended to so that kids really have the ability, the muscle to think, to solve problems, to apply knowledge. That is the core responsibility of us as an education system, public or private. And for me, 90% of kids go to public schools. That's our focus. So that if you hear me in terms of this sense of oh my God, look at what is happening both in terms of screens and then AI making it worse. Because if you are a prompt away from just lots of knowledge, then how do you actually assess whether kids have learned anything? So that was really the foundation that propelled me to the speech. And the speech then becomes really understandable because of that. And that's why the first three items were what do you ban or what do you try to pull back from in order to be able to do the active learning, the applied learning, the learning how to think, the engagement and the agency that kids need. And then what are the other things that we have to do around that, including making sure that there's privacy standards, making sure that the tech companies are actually paying for this disruption, making sure that there's resources for public education. So that's why we said ban, except for compelling reasons. Screens K through 2. Stop with actually having child facing AI or AI in children's education in, you know, in, in elementary schools, everything else with AI due under the tutelage of a teacher, except we should never have these social companion chat boxes. Let's then also really replace a lot of the very much focused simply on acquiring knowledge with active learning. And by doing that.

[B]

Randy, could you define active learning for us?

[A]

Sure. Active learning Means project based learning, experiential learning, learning by doing, meaning the kinds of things. So it could be a debate even in a classroom. Instead of actually just having direct instruction. You have like 30 minutes of, you know, four or five teams. Let's understand the facts. I mean, this is what I did in one of my classes, you know, should the US have used the atomic bomb in Japan in 1945? You know, and you, you ensure that kids have prepared for that debate. You create teams, you put kids, you, you do this in some ways on a random basis so nobody can say, oh, I'm on this side, this is the side I want to do. You really ensure that people are looking at things from different sides and then you actually have a debate about it where you ask follow up questions. So what you're doing is you're helping kids learn not only what the information is, but how to apply it, how to use it, how to work with each other. And so that's one example, but there are many others. Career tech ed is a perfect example of project based instruction. When you end up having to do a project as your end project, say it is, for example, if you are in a, you know, in a carpentry class, you are actually, you know, constructing the outside of a house. You're actually constructing, you know, a closet. Like so you have to do the various different steps and plan the various different things to get to a project. If you're in a music class, you're a composition. If you're in an art class, something that you make. So that's what I mean by project based. It's something where it's not simply giving an answer in class or writing an essay. It is something where you have to take several different steps from beginning to end.

[B]

And am I right that this is quite different from how students are using AI right now, where they are sort of reproducing knowledge, for example, getting help from an AI to write an essay or getting help from an AI to answer questions. How are students in general using AI right now?

[A]

Well, it's very interesting because I would say that nobody really knows the answer to that question because there are two new surveys out where teachers self report that 2/3 of districts, they have no idea what or they have very limited understanding of what the district policy is in terms of AI. So what's happening is that it is the wild wild west. So I've seen AI used. I went to, before I gave this speech, I wanted to go to some schools where AI was used. So I saw some Khan Academy uses of AI. And there's a very. It was used for reinforcement. It was used as. As part of a classroom, as part of a seventh grade classroom, an eighth grade classroom, in both math skills as well as science skills. And you could see it was used in a reinforcing way to ensure that kids had the information. But the kids, you, you. The kids were constantly being prompted by a screen. And that's how it was used there. I've seen it used well in career tech ed when it is used as a research tool to help ensure that the measurements are right. Like in an electricity class or in that same kind of class that I was talking about, about carpentry, like where you would see a graphic design. I've seen graphic design classes where it's used. But what's happened is that you have, in early ed, you have Chromebooks that have Gemini in it, and people have no idea. Parents have not given permission to have AI in a Chromebook in front of a student. And it's hard to know how the AI is being used or not used because it's kind of creeped in, crept in. Excuse me.

[B]

Yeah. And so how are students using AI in bad ways? What are some ways in which AI is used that undermines what we want from education?

[A]

Well, what I think is going on is, look, as you said earlier, I think what's happening is, say, somebody has a homework assignment. Over 50% of teachers have recently said, again in these surveys, that they think that students are using AI to. To replace their own knowledge as a substitute. So if they're writing an essay, AI has written their essay for them instead of them writing it themselves. And there is now a new term for this, which is called cognitive offloading, which is essentially that a student uses a prompt and AI, or a person uses a prompt and AI to gain information as opposed to figuring it out themselves. And frankly, for adults who say, you know, in. In whatever subject matter we are. So for me, say it's social studies or it's, you know, union contracts, or it's collective bargaining. These are things that I have done for my life. So I am very versed. And in, in how I would teach, for example, the Lockean Social contract. I know what it is, I know how to teach it. I know what my aim would be in terms of what I would want students to know. So say I wanted to try a new, different approach to it. And I would go to Claude, or I would go to, you know, chatgpt, or I would go to Copilot, and I would I would, you know, basically prompt them, you know, give me some information on the locking and social contract to think about and to look at how AI would actually talk about it and use it. And then I would say from my own experience, oh, this might be interesting. Oh, this is a different angle. Maybe I didn't use Locke in his historical time as to why he got to this theory. So as an adult who's very, very versed in the topic AI, cognitive offloading or having this kind of support at your fingertips could actually be helpful. But if you have no information about the Lockean social contract and you never have understood it or understand what it's about and you go to that prompt, where's the learning? And that's the problem.

[B]

Yeah, it seems that it is now very easy to basically produce credibly sounding text on any topic. And yeah, basically it's as you say, it is just a prom the way. What's the response there? Do we need to go back to having exams in person? Do we need pen and paper? Do we need to sort of have no technology? Or is there a middle way here?

[A]

I am very partial to the pen. I like the felt tip pen, so I don't. Or the Sharpie, so I don't have to, you know, so my hand doesn't hurt. But I think that, let me just say this, technology is not going away and technology is important in our lives. So this is not about just turning the clock back. But I do think it's really important in terms of teaching and learning that we help kids actually develop the habits of learning, the habits of thinking, the habits of context, of asking questions, of understanding, knowledge, of applying knowledge. Not just reciting and not just actually memorizing, reciting or handing something in where you haven't even memorized it because you have. No, because you know, you've just let the prompt do the work. And so we have to think about if we care about our kids and care about their education and care about their futures. We have to figure out ways to do some of the old fashioned things like teaching with paper and pencil, with paper and pen or paper and pencil. We have to also think about sometimes doing exams with the old blue books instead of. And we have to do some of this work. Does it mean just doing everything in the analog world instead of the digital world? No. Doesn't mean doing everything with paper and pencil?

[B]

No.

[A]

But this is really important to understand brain research and the process of learning and ensuring that our kids are learning and they are learning how to think and feel Some facility with how to think.

[B]

Is there a risk that if kids never learn, say they use AI from an early age and they never learn how to learn, they never learn how to think. Think, what situation do we end up in?

[A]

It's terrible. I think that we're. I think that it's not simply what we just saw in terms of the decrease in test scores. It's. We're not helping our kids. We're not teaching them how to learn. We have to be lifelong learners. So, you know, I fear the prospect of that, which is why, look, the speech is kind of screaming from the rooftops. We must do something now. From an educator who really believes in active learning. And my members are. I talked repeatedly to leaders over the course of this year about what we were seeing and what we've seen. In the aftermath of the speech, I did some briefing of some parents beforehand. I've only gotten positive feedback in a time when literally anything I say is met in the culture wars with lots of critique. Let me just say that this time it has been overwhelmingly positive. And my criticism, I'm very critical of the Trump administration and of Linda McMahon in particular, because this is her job. I mean, you don't. The Department of Education doesn't run schools, but in. In something where the future is at stake of our kids and, and the administration has lots of different policies about AI for you, for the Department of Education, for the Secretary of Education not to actually give any kind of guidance or feedback to people, to basically do. To basically simply say, well, they should use it appropriately. Well, what is appropriate in this moment in time for the acting Surgeon General to be the only one in this administration to actually say, oops, we have some fear about what's happening with kids, and maybe they should use paper and pen again. It's really ridiculous. We can't have. This is a huge change in society. I don't care who the government is. I don't care if they're Republican, I don't care if they're Democrat. I don't care who they are. They have a responsibility to try to get to the best guidance and the best research, and they're doing none of it. And that's part of the reason why I called for a research consortium to basically say, look, let's deal with the harms. Let's deal with what we need to do going forward. It shouldn't simply be Jonathan Haidt's research or Jared Cony Horvath's research that everybody is relying on from books that they both wrote.

[B]

What does this consortium look like, because you mentioned that it shouldn't be funded by industry, it shouldn't be funded by industry. Basically it should be independent. Who should govern it? What are the biggest unanswered questions in this space?

[A]

Well, there are so many big unanswered questions because nobody has yet stood up to say, I will do it. I mean, I think some of the big philanthropies, like they funded other things. Some of the big universities, remember in America we have had research consortiums in healthcare, in health research, between some of the biggest universities, some of the biggest teaching hospitals, and the government. And ultimately that's what it should be. The fear is you don't want to have the situation that happened with tobacco, that the tobacco industry was able to basically hide information and data for years and years and years until in some ways it was too late. So we need good research. I spent a lot of time looking at the research that is available, talking to teachers, talking to kids, teaching myself, seeing what active learning does in terms of its potential and seeing what cognitive offloading or being a prompt away to have information. So information is at a kid's fingertips without learning it, or at an adult's fingertips without learning it. And also seeing the impact of screenshots. I mean, if, if, if, if you can't blame kids if the screen is interesting and it's fun and it's engaging and, you know, learning takes some work. It's not a surprise that somebody would say, oh, let me get the immediate gratification. So, so that's part of the reason why I think in schooling we have to make schooling interesting, but we have to be able to teach kids how to learn.

[B]

Yeah. One concrete suggestion you make is a ban on screens up to and including the second grade. Is there a specific reason why younger children are more vulnerable to basically the enticing effects of screens?

[A]

There's two reasons. One, they don't have the maturity. Number two, their minds are more supple than any other time in our lives. So that's when so much learning occurs. That's when the habits of learning occur. And there's lots of research that shows that's why so many people talk about early childhood and how important it is. There's so much research about parents reading to kids. Kids have more knowledge about words and vocabulary than parents that don't read to kids. And in fact, one of the things we do as a union is we have a program called Reading Opens the World. And we have basically spent somewhere between, you know, somewhere around a million dollars a year for the last several years to give books away to parents and kids, you know, in community events and other things. We've given over 11 million books away to parents and kids to, to spark the love of reading.

[B]

Is there a reason why it shouldn't go further? So say third and fourth and fifth grade, why specifically the second grade?

[A]

Because I. Because early childhood is different than. That's why I strive to say that in that, that in early childhood we really have to divorce ourselves from screens unless there's a compelling reason. So that kind of teaching and learning that you do in the physical space really happens. The play really happens. Kids learning how to have relationships with each other, learning the skill of persistence, learning the skill of attention, all of that really helps with cognition. But in then the elementary school grades of third, fourth, fifth, sometimes it's sixth, sometimes it's seventh. But in the elementary school grades, really trying to think about what it would mean to teach without AI. Student facing AI. Now I'm not saying, I'm not suggesting, when I say student facing AI, it's different than teachers using AI as a tool. And many times these things get conflated together because teachers using AI as a tool, administration using AI as a tool is really different than students using AI. If you don't have the safety and the privacy and if parents haven't had some input into this, then there needs to really be lots and lots of different protocols. That's why as students get older, you can do a lot more digital literacy, which is what is actually needed. The ethics of AI, the using it safely, the limits of it, misinformation, disinformation, all of this should be part of a digital literacy package that teachers and parents know before there is real use of it. I mean, let me say it this way. Driving a car, would anybody be allowed to drive a car without them having the training about how to do it? So in the physical space we take the precautions we need to take, but in the virtual space we do not.

[B]

Yeah. Another suggestion you make is that children under 16 shouldn't interact with social chatbots. They shouldn't. Yeah. They shouldn't learn from these AI companions. Why, why is that? What's the dangers here?

[A]

Well, the social chat bots. So much of what is happening right now, and you know, I'll go back to the well being piece. One of the things that technology and social media look Covid made it worse because of the pandemic is that you see research after research, people are alone more. They don't engage with Relationships as much they don't hang out with each other as much. So what happens? That chatbot becomes your companion. So when a chatbot, a robot, and something that is not an entity, it's programmed. When it keeps telling you I love you, it keeps telling you you're fine, it keeps being sycophantic in a way that you then are addicted to it. How does. How is that helping a child? And in fact, if a child is so vulnerable or young adult is so vulnerable, and every young adult is vulnerable, Every young adult, all of our hormones, when we are in our adolescence, we're all vulnerable. And so how does that. I mean, it's not just the terrible stories we've heard about kids who have suicided. How does that help someone break into the real world and have the skills and knowledge and the sense of self and confidence and persistence and tenacity to deal with the real world? And so, okay, when somebody's more of an adult. This is the United States of America. We believe in freedom. But kids, we have a responsibility.

[B]

What is it that worries you about AI simulating human relationships?

[A]

Everything. Literally everything. Because it doesn't simulate human relationships. It simulates syncofancy. It simulates hearing what you want to hear. It simulates just making you feel okay. Now, I want kids to feel okay all the time, but it doesn't give you any resiliency or any sense. And it makes a premium. It says that the isolation is okay. And so for a chatbot to tell you that it loves you, think about the propaganda that a chatbot could then be programmed to do.

[B]

Yeah. You can probably form relationships with AIs that then leave you quite vulnerable to being convinced of all sorts of things, including buying products or what you can imagine. Yeah. Bad things happening here.

[A]

It's not simply. It's the. To see that you can form an emotional or an intellectual relationship with an inanimate object, then what? And that inanimate object is constantly. Then you create trust, and then it's constantly in your ear about anything. So when will we stop it? Well, will stop it when the inanimate object starts telling people to kill each other. I mean, so what's the line? And so for me, the line is, let somebody have the emotional maturity to actually make decisions. That's why I said 16.

[B]

Yeah. You talk in your speech about gold standards for AI safety and privacy. What would standard. What would those standards look like?

[A]

So what? You know, we are in the. One of the things we have done this year, you know, kind of this year for The AFT has been bookended by. At the beginning of the school year. At last August, we announced an AI institute which is being funded by some of the AI tech companies to actually teach teachers about the tool. Now they're not doing the teaching, we are doing the teaching, but they helped us fund it. And over 12,000 people have gone through this AI institute already in terms of understanding and using the tool. By the way, one of the courses that is most requested is one about cognitive offloading and how we teach critical thinking in the midst of this AI world. The most sought after course. So the. So, so we, but we said to the three companies we're working with, Microsoft, Anthropic and OpenAI, we said that we need to have a gold standard, something that doesn't exist right now in America. It should exist in law, it does not. But we need a standard to ensure to parents that your kids, your kids data is protected. We need to make sure the teacher's data is protected, that it can't simply be anytime someone in school opens up a computer with AI in it, or a laptop or a cell phone, that the company who runs that particular AI has all that data. The parent didn't give that permission. And you know, and one of the most interesting conversations I've had, we've had with the companies is when they say, well, you know, school districts never asked us for this. Well, school districts should have asked you for this. But part of it is they didn't know enough to ask you for this. But this is so we need to have this kind of. Parents need to be in control of the data that is available for tech companies to use from their children.

[B]

Are you worried about the companies using that data to train further models, sort of learning about the preferences of teachers and students?

[A]

Yes, totally. I'm totally worried. And that's what has happened with social media. And that is what has happened. Because what happens, AI learns from more AI from more data. So if data is regularly accessible on the, in the public, it's one thing, but if it is something that is. So this is what we are struggling with. We are struggling with what is the permission structure, what is the way. And we believe in transparency and we believe that there has to be that kind of permission structure. So we are in the midst, we're having good faith negotiations. We're in this kind of 60 day period where we're really trying to wrestle it down and hopefully I'm optimistic that we'll get to something. But I've also said to the companies that if we don't get to something satisfactory. I feel like we have to walk away from the institute and they don't want to do that and I don't want to do that. But I feel very. This again, I'm going to say this over and over again. I come to this from being with teachers, from watching kids struggle, from wanting kids to have their future and have their American dream, and to see that the screens have really harmed kids and their ability to think and to attend to schoolwork. And AI has the potential of basically being a prompt away from any kind of knowledge, so that you never know whether a child has actually learned something or not. And that's why it is incumbent, I think, on all of us to actually focus on what we need to do to right this ship, to make sure that the balance is there so that teaching and learning can really happen for kids in this era. And frankly, this is the irony. In an AI world, kids really need to be prepared with critical thinking and they really need to be prepared with understanding context, with understanding what's going on around them. They really need to be able to think and problem solve and do higher order skills. And so what we're doing is we're actually preparing kids by saying, take some of this away right now, Stop with the easy stuff, stop with the attention shift and really let's figure out how to teach kids so that they are prepared for this world. That's the irony here. You got to take some stuff away to prepare kids to think in order for them to be ready for this world.

[B]

Because the easy solution of sort of cognitive offloading to the AI leaves you not as skilled in the future or not prepared for future facing the actual world.

[A]

No. And you know, I used to call it, how do we know whether kids actually have knowledge? How do we know whether they've acquired it? And then, you know, the AI world came up with a term that defines that cognitive offloading.

[B]

Which decisions should remain with individual teachers or individual schools, and which decisions should be made on a higher level. So should there, how much room for implementation should there be with AI?

[A]

Well, there should always be lots and lots of room for implementation on a school level and frankly on a classroom level. But if we don't have some of these broad guidelines, then it's hard to do any of it. Because right now it's a wild, wild west. I mean, there, there really needs to be like, I'll give you an example right now in New York City, because of the work that we've done with this AI Institute, teachers are Coming up with really amazing ideas for how you use AI, how they use it in their lesson planning, how they use it, how they're thinking about using it to help them plan, to help them teach. But then one of the issues that have come up is the writing of IEPs. What happens if we actually had an AI template to, in terms of IEPs, it would be a huge time saver for teachers.

[B]

Can you define that term?

[A]

Randy, IEP is an individual education program, not the specifics of everything for a child. But there's such a backlog in the writing of IEPs, in the needs, in providing the needs and services to special ed kids. So when thinking about should that, you know, can we come up with a format, can we come up with something that works, can we save people time? Can we still make sure that we have the personal attention to it? So that's something that has come up in the context of our AI Institute, just talking to teachers of kids with special needs and what their needs are. So there's a lot, there's a lot that people are going to be able to learn from. So you need to have ways of ensuring that people are learning from each other. You need to have transparency, you need to have good communication. We'll learn a lot from each other in the next few years that we don't know right now. Like, if you had asked me two years ago, what could AI do, I would say, well, probably people could use it to help them refine their lessons. But you can see it can be used in so many ways in instruction right now. If we actually had the guardrails in place, if we actually had safety and privacy and guardrails around, ethical usage and guardrails around, you know, is this a higher order skill or is this just acquiring information? So, you know, you just need, you need a lot of that. What happens in terms of higher ed teachers, their own, the intellectual property, the research assignments that they give, the research they themselves do, and how without that them having a right to own their own research, how AI then scrapes that research and then it goes into the broader domain and they've lost the right to their own research. So there's a lot of issues here that really require some degree of regulation in order for people's safety, economic security to be protected. And right now what's happened is the balance is totally off. The tech companies have all the rights and they have all the, basically the apparatus and individuals don't have virtually anything. So in that context, if we had something like no screens for kindergarten through Two and except for compelling reasons and there was like a definition of what compelling reasons are, then what would end up happening is that people would make some decisions in K through two classrooms. Maybe a teacher had, maybe a school district like New York City, which has virtual days for snow days. Same with a lot of places up north because of snow days. Maybe that's an exception. Maybe an exception, you know, in the south would be if there's hurricanes or things like that. So I'm saying that there has to be some elasticity here. But the key is, does somebody like la, do you believe that the screens have been harmful enough that we got to actually get kids into teaching and learning absent to screens in the earliest grades?

[B]

Yeah. When you talk to the AI companies, how do you think they are thinking about AI? Where do you disagree and where do you agree?

[A]

Well, I don't talk to all of them. There are some that A wouldn't talk to me and there's some I wouldn't talk to. I mean, you know, so I don't, you know, I, for example, I got off of X. I mean I'll still read what's on X, but I do not. Neither I nor the union posts on X anymore because you know, I thought that it was one too many abuse of children scandals that they did not, that Musk did not deal with and Grok did not deal with. So you know, there's no, there's no, there's no conversation between that AI company and us. Same in terms of Peter Thiel. But the companies that we talk to, you know, we are saying to them, it can't be profits before people that you're in a business that is actually being used in so many ways by people. You know, all of the chatbots are subscription based. So we have to actually make sure, just like with the physical world, that not just that team humanity is in charge of the world still, but that team humanity is very much part and parcel of the product that they're selling. That's why safety and privacy are so important to us.

[B]

Say your ideas are implemented. What does the education system look like in 10 years?

[A]

Much more vibrant than it is right now. And if you're actually doing so much more with active learning, if we're actually understanding that things like music and art and sports and debate and career tech ed projects and a project, a science project, if we understand that these kind of things that we see as either end term capstone projects or this is the way in which classrooms are organized, then I think you'll see a lot more vibrancy. And more importantly, you'll see a lot more agency for children, that they will feel more responsible for their education. They'll feel more excited about it, more engaged in it and more confident about it. And I think teachers would like it a whole lot more.

[B]

Randy, thanks for chatting with me. It's been great.

[A]

Thank you.