AI Honesty in Education: The Need for Truth

This transcript was generated using AI and has been reviewed by humans for accuracy. Minor errors or artifacts may remain.

Vicki Davis (00:00): Welcome back, educator. Today we’re going to have some honest conversations about artificial intelligence, something we all need to be having.

Announcer (00:09): Ever wondered how remarkable teaching happens? Find out right now at Cool Cat Teacher Talk with award winning teacher Vicki Davis. Get insights from top educators, tech tips, and inspiration to elevate your teaching.

Vicki Davis (00:21): I have a story I like to tell students that I call the Honesty Telescope. This story is from Benno Mueller Hill, a professor at University of Cologne at the Institute of Genetics, who told this story in September 1993 edition in the Quarterly Review of Biology. This is a story from when he was in high school. Now I want to know. I wrote this story down around 1995 to put in my quote box. Benno said that one morning in high school, he was last in a line of 40 students waiting to look through a telescope.

This telescope had been set up by his physics professor to view a planet and its moon. Now, the first student stepped up and looked, so the teacher asked him if he could see in. The first student shook his head and said, no, I’m nearsighted. So the teacher showed him how to adjust the lenses and turned the knobs to focus it. After lots of adjusting and frustration. The boy finally said he could see the planet in the moon. Every other student said they could see it right away.

Every student just saw what he was supposed to see. Planet in moons. But finally they got to the second to last student. Student. Number 39. Harter, Harter looked into the telescope and said, I can’t see a thing. So the teacher shouted at it, you idiot! You have to adjust the lenses. The student tried but said one last time. I still can’t see anything. It’s all black. Then it says that the teacher was exasperated and finally looked the telescope himself with an odd look.

The lens cap still covered the telescope. Nobody, none of those 38 students had actually seen a thing. They all said that they could. Benno asked himself if he would have had the courage, as the last boy in line, to admit that he did not see anything. So when you’re approaching the honesty telescope, it means that even if everybody else has looked at something, look at it yourself with fresh eyes. And when you do, you need to be honest about what we actually see, when we feel pressure to see what everybody else is seeing and don’t share the truth of what we see.

It makes it difficult to learn, grow, and have honest conversations. I know there’s a pressure to agree with everybody else around us. People who say that that AI is terrible or that AI is great. But often those answers are actually in between. And there’s good and bad uses for every technology, just like their good and bad uses for educational technology, some that improve learning and some that take away. So when we talk about artificial intelligence, so much is unknown.

First we’re going to talk to Justin Reich from MIT. And we’ll learn how in the age of AI, in the absence of extensive research studies, which we may not have for many years, we’re going to have to be willing to create micro experiments in our own classrooms, in schools. Let’s talk to Justin.

Announcer (03:42): Cool Cat Teacher Talk with award winning teacher Vicki Davis.

Vicki Davis (03:46): Today we’re so glad to welcome back my friend Justin Reich. Now, Justin is actually one of those who saved my classroom in real life. He is an associate professor. Of digital media at MIT, the director of the Teaching Systems Lab. He’s the author of Iterate and Failure to Disrupt. And he’s back to talk about the homework Machine. His brand new limited series podcast that dives into it, AI is really doing in our K-12 classroom. Based on 120 interviews with teachers and students across the country.

Justin, you’ve gone and conducted these 120 interviews about AI in classrooms. What made you think that you needed to, like, get the real story directly from teachers and students?

Justin Reich (04:36): It was the almost exact same motivation that had me visit your room 20 years ago. 20 years ago, all kinds of folks were talking about web 2.0 in schools. A thing that I understand better now is that when new technologies come along, elites dominate the conversation. Think tank people and self-described thought leaders and policymaker kinds of folks and things like me. And I really don’t actually trust any of those people. I trust classroom teachers and students a lot.

At the very least, their voices are essential. So for the same reason that I wanted to visit your classroom and see what was really happening in your environment in Georgia 20 years ago, I wanted to say, all right, you know, ChatGPT has come and crashed the party. It has showed up uninvited in all of these different schools, and teachers and students are just bringing it into the classroom on their phones. And what do they think? And what do they say about it?

Because, you know, sort of professors and thought leaders can think whatever they want. But the most important observations are the ones from the people who are closest to what’s actually happening.

Vicki Davis (05:37): Let’s kind of back up at the 30,000ft view. But what kind of conclusions are you starting to draw?

Justin Reich (05:42): We’ve had a hard time concluding, because there’s 130,000 public schools in this country. There are 13,000 school districts and it hits different places really differently. There are lots of schools in the country where there are no substitute teachers and kids come to school. Hungry kids are not showing up to school because of chronic absenteeism and, you know, really huge challenges. And in those kinds of places, AI tended to be described as the fifth, 12th, 19th thing on people’s lists.

Like, if kids don’t show up to school, it doesn’t really matter what’s going on with AI. When teachers said to us, I would love to do a day of professional development on AI, there are no subs. There’s no one who can come into my classroom and have me leave. It tended to be in more affluent places where people said, like, this is the number one concern. This is the thing that we’re really tackling, and then people just have wildly divergent opinions about what’s going on.

There are some folks who said, like, this is a complete game changer for my classroom. I’m super excited about what’s happening. I’m thrilled about what my students can do. And there are other folks who said, this is a machine that put words in my students mouth that aren’t their words, and how am I supposed to teach someone if I’m just getting words from a machine and not from my actual students? What’s this going to do to trust? What’s this going to do to our community?

Really wide ranging opinions. Probably some of the most exciting stories are where those wide ranging opinions are in one community. I’m sure there’s some of that in your school. I’m sure there’s some of that and all your listeners schools and just hearing about how communities are trying to negotiate these challenges on a time scale that nobody asked for. Nobody gets to pick. This is the AI year. It just all arrived at once. Part of what we’re trying to tell schools and district leaders is like, you actually do have to deal with it.

We’re not exactly sure what you’re supposed to do, but you have to do something and you have to do something together. Because if you let each individual teacher figure things out in their own classroom, that’s the real recipe for chaos.

Vicki Davis (07:35): So if you have one teacher like me, actively use an AI. I teach kids how to do presentations with AI, create their presentations, the spreadsheet where my kids are writing lookup tables that I never could have had. Eighth graders. Right? Right. All these different things. But then maybe somebody down the hall thinks it’s, quote, cheating to write a presentation. I tried to make sure, is this the student giving the presentation or are they reading their slides?

But here’s the deal. Ten years ago, if they read their slides, they couldn’t make a good grade because it wasn’t their presentation, because their mom might have written it right. So you’re saying that that when we take all these different approaches that it really can harm learning?

Justin Reich (08:13): I think the most important thing is the community talking about it together and coming up with some shared expectations, because you could actually imagine a really good school curriculum where imagine your school, some of your students head to Miss Vicki’s classroom, and they’re getting this really AI intensive experience. And those same students in some other year are going into a classroom where they say, we’re actually not going to use a lot of this stuff.

There are a lot of techniques that humans have developed over thousands of years for communication, and we’re just going to focus on those because it’s good to get good at those two. And a student has a sort of coherent experience through this thing where they go, oh, sometimes we’re doing more traditional things and sometimes we’re doing newer things, and I know which is which, and that’s okay. I think the real chaos is when 10th graders get spread across, are in Miss Vicki’s classroom and some are in somebody else’s classroom, and Miss Vicki is expecting this, and the other person expecting this, and they’re not getting the same experience.

And there’s not the same set of rules. It’s coherent working together amongst the faculty to say, okay, we’re all going to do it this way, or we’re all going to be really intentional and communicate to students how we’re going to do it differently. You know, one thing that we’re pretty sure of from the state of the research right now, is that we don’t know the best way to do this. This is not a moment to sort of look for best practices because the best practices aren’t developed yet.

It’s a moment to experiment, but the way that you help students feel like they’re part of a meaningful, coherent set of experiments is that you tell them that that’s going on, and you talk across faculty, so you feel like whatever it is that you’re deciding to do, you’re doing it together with some sort of joint set of expectations.

The main advice that we have for people is create spaces where faculty and talk to each other and whatever they decide to do, because there’s a lot of potentially good approaches to do it together and to help students see that they’re part of a community of people who are feeling their way and systematically experimenting their way through this thing, as opposed to just there’s this set of rules in this classroom and this set of the rules of this other classroom, and nobody knows what the connective tissue is between any of it.

Vicki Davis (10:10): So let’s talk about this research for a minute. I just did a piece of my newsletter, the Stanford Meta Research, where they look all the places studying AI, and they found only 20 of them had any measurable results. But then none of them are in the basically US K-12 classroom. And it seems like to me there are a lot of people trying to draw far reaching conclusions from research that’s in its infancy. Do you see that happening?

Justin Reich (10:34): It is exactly right. Two reporters in the past week, one from the New York Times, one from Ed Week, saying, like, what are best practices with AI right now? And I hear that from teachers all the time. Like, what are the research based best practices? That’s a really good intuition for teachers to have in lots of things, like, if your students are having a hard time reading, you should not invent reading instruction like we’ve studied teaching reading for 60 years, and we can tell you better and worse ways to teach reading.

We can’t do that yet with AI. To give you a kind of benchmark, I think Google was founded as a company around 1995, first peer reviewed paper that had really solid research about how, like the most effective ways of teaching kids to sort truth from fiction on the web was published in 2019. So it took about 25 years for the research community to say, we’re pretty sure this is what you should teach students to do when they’re using a search engine to find facts, the arc of time that it takes for the research community to come up with, like, pretty solid answers to important questions is unfortunately, like closer to decades than years.

Ten years from now, you and I will have this conversation and we won’t be like, what should I policy be? We’ll be like, oh, we’ve studied like AI policy in schools for a while, and we’re pretty sure that when schools do this kind of thing, it doesn’t work as well. And when they do this kind of thing, it works better. But we’re actually still kind of a long way away from that. When big science is taking a long time, then what educators needs to substitute is local science is going into their own communities and saying, we don’t have all the right answers.

So what we’re going to do is an experiment, like the best ways to conduct experiments or to a tell the people involved that you’re experimental. So parents and students and teachers should know these are just things we’re trying. There are no best practices yet. This is our best intuition of the way to go forward. And then you evaluate the evidence afterwards. So, you know, you were just telling me a story about having your students do speeches in class, and you’ve had students do speeches in your class for decades.

You’re saying, oh, when we do this AI enhanced approach, the speeches were better. You know, I graded them. I compared the grades from 2026 with the kind of grades I got in 2019. And because the performance of understanding is better, I have evidence that the thing that I’m doing is working. You could imagine there are other experiments that you could do where you try an AI enhance thing and you’re like, oh no, that made it worse. And you say, okay, that’s a bad experiment, that one we’re going to throw away.

But that, I think, is the crucial stage that we’re at that sort of local educators with their colleagues conducting their own local classroom experiments in this period of uncertainty, where the research summaries that you’re going to get for the next decade are not going to give you the sort of slam dunk answer, because big science just takes longer than that.

Vicki Davis (13:20): I’ve been using Claude Cowork a lot and I’ve written a skill to take the research on adding fun to learning that I run my lesson plans through so that it gives me more ideas on how to have fun in the classroom. It gives me ten. I decide which one to take. Oh, we gave the worst speech contest. So the kids had a checklist, and their goal was to have as many wrong ways to give speech as possible. It was hilarious. It was wonderful. And truthfully, they made better grades now than they made last semester.

Justin Reich (13:50): I think there are a few things there. When people use tools to make their work better and make their students learn more. To me, that’s winning. So a thing that I really like about your example is that you have this clear vision of what high quality work looks like, honed over two decades. And you’re saying, look, I got this tool. And it helped me get better learning outcomes from my students. So anytime teachers, you’re talking in that kind of way of I did a thing before I was clear about what good outcomes looked like.

I tried a new thing. I got better outcomes like that, to me, sounds like winning. And it doesn’t matter to me whether you’re winning with with AI or cereal boxes or whatever else. I think educators should be conscientious that students do bring a deep suspicion about teachers using AI potentially inappropriately. We got lots of stories and examples of teachers starts getting tired on a Sunday night, and they have AI give too much help making a quiz, and some of the quiz questions like don’t make sense and they don’t work anymore.

And the students take the quiz and it’s messed up and they go, yeah, you know, you were making stuff up with AI and I think rightly so. They become quite suspicious. Like if you’re not putting in the effort, why should I put in the effort? Perhaps the hardest thing about your story is you. You had an important detail, which is you said I asked AI to give me ten suggestions and I picked the good one. You can pick the good one because you’re a master teacher who has been doing this for decades.

When brand new teachers or any not. I mean, the thing about AI is that it’s utility depends upon your domain knowledge, but they’re novice teachers who have never taught before, who are looking at ten suggestions from a lesson plan suggestion machine of any kind. They’re new, and so they can’t tell the good ones from the bad ones. Just like when we ask our students to use AI to give them suggestions on their writing or something like that. If they’re novice writers, they can’t actually tell the difference between the good suggestions and the bad suggestions.

Vicki Davis (15:54): About domain knowledge. How would you define domain knowledge?

Justin Reich (15:57): In progressive education? There’s sort of often a hope of something like what we’re going to teach students critical thinking. We’re going to teach students 21st century skills. We’re going to teach students these like big, broad ideas that are applicable to lots of things. And it actually turns out that being an expert and being creative and being good at stuff often involves like having lots and lots of facts stored in your head. Probably the best research we have on this is about chess.

If you teach students to become masters of chess, they’re actually not any better at anything else. They’re not better strategic thinkers, they’re not better creative thinkers. And the way that we make people be really good at chess is basically having them memorize historical chess games. Like, if you want to become a grand master at chess. You have to play a bunch of games of chess. But then the other thing you do is you just read these books, which are about other people’s chess games, and then over time you start to like, recognize patterns in chess.

But, you know, it’s basically like storing a series of facts in your head. It’s kind of weird to think that Magnus Carlsen is way better at chess than you or I, because he’s memorized a bunch of chess facts, but like, the reality is pretty close to that. And so similarly, like if you want to be a good plumber, you actually have memorized a bunch of plumbing facts, like some of those plumbing facts are about, like how you turn things and move your hands and things like that.

So I have memorized lots and lots of history of education facts and history of education, technology facts. And so when I ask to help do research or to do. Education related things, it’s pretty easy for me to spot when they’re wrong, because I know lots of things. When my novice students who are MIT students who are brilliant in many ways and know lots more than me in lots of areas all the time, they write papers where they’ve used GPT to summarize the research or other things like that, and it’s pretty obvious to me that they’re totally wrong.

And it’s not because they’re stupid kids, they’re incredibly smart kids because they have no domain expertise in my class. That’s why they’re taking my class. I think the advice that you’re giving students is, is exactly right, that in the areas in which we really know a lot of stuff, we have some pretty powerful tools to discern the quality of GPT output in areas in which we don’t know a lot of stuff. We really are much more limited in our capacity to look at a set of output and say, that’s true, that has good evidence behind it.

That’s creative, that’s interesting. We have an episode of this in the Homework Machine, where my colleague gets GPT to write a song in the style of Johnny Cash, and he does a great Johnny Cash impression. And so he asks GPT to come up with a bunch of, like, potential song verses. And most of them are terrible, but a handful of them are really good and really funny. And the way that Jesse knows the difference is he’s a songwriter. He’s spent decades trying to write music, listening to lots and lots of music, memorizing song lyrics.

You got all these song facts in your heads. One of the good things about that if you run schools is that, you know, for a couple hundred years in the United States, schools have been in the business of developing domain knowledge. Like the thing that we do is we help people make sense of biology and physics and social studies in US history and world history and British literature and whatever else we’ve decided to do. The if it were the case that the thing that made us good at AI was some sort of specialized set of AI skills, then we’d be like, man, we really got to change the curriculum here, and we’ve got to teach all these new and different things.

If the thing that makes us really good as humans at partnering with AI is deep domain knowledge, then like, actually, maybe we don’t have to change schools that much at all. Like if we want students to be really good microbiologists using AI to invent new proteins and drugs or things like that, then maybe the thing that they really need is a really good education in biology so that they can discern outputs. We’ll still have to learn more in the decades ahead about what kinds of things help people be the best partner with GPT and AI and whatever else gets invented.

But if I was a betting man, I would put a lot of bets on domain knowledge, and in some ways, it’s good news for schools, because that’s one of the things that at our best, we’re good at.

Vicki Davis (20:00): So you’re called the show The Homework Machine. And isn’t that how a lot of students are viewing AI? It’s like, hey, the teacher that says, answer the questions. At the end of the chapter, I want to take a picture of the chapter and take a picture of those questions and be done. What do we do about this? Does that mean homework goes away?

Justin Reich (20:19): My favorite thing that students will tell us is they’ll say things like, oh, I don’t use AI for homework. And then you keep talking to them for a while and they describe using AI for homework and you’re like, And they’re like, oh no, that’s busywork. You know, a lot of students will decide, like, this is the homework that’s not important and doesn’t count. And this is the homework that is important. It does count my teacher head. I’m thinking to myself, like, you don’t get to decide that.

Like I’m not. I’m sure not all my homework is perfect, but I really only assigned you this because I really believe that need to be needed to be in your head for you to be able to do other things in the future. So you know for sure lots of students are using GPT to bypass learning. I mean, probably our favorite episode of the Homework Machine is episode four called busted, which is interviews with five students about how and why they cheat, some of the stories about them hitting a wall.

You know, kids getting late at night and being like, I’m just going to have the machine do it. We have a great interview with one student who just did homework, all of his homework his senior year. He just had GPT do for him. And to some extent, my interpretation was it was just kind of boundary pushing. It was just like, I’m an adolescent. My job is to figure out what I can get away with. Nobody stopped me, and so I kept getting away with it. You know, one of our favorite stories is from a student who really loved her International Baccalaureate Theory of Knowledge class.

And it was taking up a ton of her time, and she wasn’t using AI to do any of that work. She was using AI to do her other classes because she really wanted to do her homework in this other. So it was it was like skipping some homework in order to do other homework. So students have a lot of different approaches. We saw three different ways teachers were dealing with this. One we called Beg and Plead, which is basically this, you know, in the moments in which you think students should be doing work with their own brains, you just say, I promise you it is a bad idea to use AI.

Please don’t do it. That tends to work best in sort of small school communities, where students trust that they’re going to get a lot of human feedback. So in places where people feel like they have really close relationships with their teachers, and their teachers are meaningfully looking at their work, that seemed to work okay. Another one was kind of like the constant, like the nag and do not accept approach where there’s not a lot of like punitive things.

It’s just people turn in work that you’re pretty sure they didn’t do and you’re like, man, do it again. You know, we had one teacher who was like, I’m the pettiest person, you know, like you’re just going to like, you keep turning this stuff in and I’m going to keep turning it back. We have a great story of a teacher who, when he gets stuff that he’s pretty sure is AI. I love this detail. He’ll take a phrase and he’ll write it on an index card. He’ll take words and he’s like, pretty sure the students don’t know and write it on the index card.

And then he’ll call the student up to the table and pass them the index card and be like, you know, hey, Jim, what does this mean? And Jim’s like, I don’t have any idea. And they okay, like because this was in your paper. And then we had, you know, their schools that are using detectors and, you know, using the detectors immediately punitive using detectors to spark a conversation that leads to punitive approaches and things like that and their justifications for that.

And then maybe the fourth approach is one that we’re called calling bog trotter, after Bruce bog Trotter and Matilda, who eats a piece of cake and then is forced to eat the entire cake where the teachers are just like, you must use AI, you’re going to use AI to cheat? Fine, then everything has to be sort of all AI all of the time, and I’m just going to make the assignments much harder so that you can’t complete them without AI. And I think as I was reflecting back in my classes, I pretty much use all of these strategies in one form or another.

We don’t know what the right approach is, but I think some of the things that do seem to matter, our colleagues talking with each other about the kinds of things that they’re going to use in their schools, and the kinds of programs that they’re not going to use in their schools, so that hopefully there’s some consistency, or at least when there’s inconsistencies in departments are greater, they’re described to students. And then are people looking at student work before and afterwards.

You know, if your plan is to just like beg and plead students not to use AI, but you’re getting all of these garbage, homogenized essays that kids are obviously cheating on, then your strategy is not working. You’ve got to, like, go back to the well and you’ve got to pick a new strategy. I think what we can do at this point is describe a range of approaches that teachers are taking, and tell some stories about where they seem to be working and where they seem not to be working.

But it’ll be a little while before anybody can say like, okay, these are the ways that we know don’t work at all. And these are the ways that we know work pretty well.

Vicki Davis (24:46): Well, my approach is when we do the coding, I pull up what they turn in and, I kind of know when it’s a little past their ability. And they have to explain every line of code. And you know, what if they wrote it with AI, but they learned what every single line did? Hey. They used AI to learn how to how to code at a higher level than I can teach. I’ll take that all day long, but anything you turn in for me, I guess it’s kind of like the index card approach, except I try not to be too.

Gotcha. It’s like everything that you turn in is the topic of a conversation. Before my students did their presentations, they had to use Google NotebookLM to do their research. And I interviewed every single one of them, and we sat down and had a conversation. And if they didn’t understand that topic, I’d say, hey, you need to go back and look at this, this and this. I’m coming back to you. We’re going to talk about it. I got to ask you one more question, though, because you wrote about subtraction in ASCD, The Power of Less, where everybody’s trying to add AI to everything.

What should we be subtracting as we finish the show?

Justin Reich (25:47): The whole idea of subtraction is that schools are too complicated today. There are too many things going on and we can’t be good at everything. And so we need to be deliberate about taking things away in the national conversation. Almost the only thing that schools across the country blue states, red states, private schools, public schools have agreed can be subtracted from schools is cell phones. And you and I could have a long conversation about whether or not cell phones belong in classrooms.

But here’s the thing that I celebrate is that people said, we’re just not going to deal with this anymore, but we’re going to put them away. And maybe they would have been a good learning thing. But here’s one fewer thing that we’re going to deal with so that we can deal with more important things. And I actually celebrate that part of the decision. But schools just have to decide. Cannot keep adding, you know, new standards, new technologies, new like schools cannot solve all the problems of society.

We have 180 days. We’re about seven hours a day. It’s actually not that much time. And so I think good exercise for schools to be regularly doing is in their cycles of professional development, improvement and things like that, or like what are some things that we can stop? What are some things that we can set aside because we want to do a really good job on a manageable number of things, not a mediocre job at an unmanageable number of things. Because our schools are so diverse, it’s really hard to say, like, this is the thing that you should definitely get rid of, but it’s the things that are just kind of limping along and not really working anymore.

Finding what to prune is the way that you get your best stuff to grow.

Vicki Davis (27:19): I love the great way to end. So Justin is an associate professor of digital media at MIT and director of the Teaching Systems Lab, author of Iterate and Failure to Disrupt. And we’ve. Been talking about the homework machine available wherever you get podcast.

Announcer (27:37): Cool Cat Teacher Talk with award winning teacher Vicki Davis.

Vicki Davis (27:41): Again, we need to think about how are we going to create experiments in our own classrooms and in our schools? Once I knew a remarkable woman named Shabbi Luthra, she was Director of Research and Development at the American School of Bombay in Mumbai, India. It was her job to conduct experiments on every tool that they used at that school. They created an environment where they were constantly piloting, testing results and deploying them to the whole school when they worked.

I like what Justin said. Finding out what to prune is how your best stuff grows, and that’s what we need to hear right now. We don’t need to do everything. We need to test things and keep only what works. I told the story of the Honesty Telescope at the beginning of this session, where 38 boys in a line of 40 looked through a telescope and claimed to see a planet and a moon. But a little boy named Harter said he couldn’t see anything. That was when they found out the lens cap had been on the telescope the whole time.

The author of that story, Benno Mueller Hill, also said in that 1993 paper where he told that story that in science and elsewhere, there are two types of truth. Number one, the truth everybody already knows. And number two, the truth that is not yet discovered. He said that most scientists simply just produce more of the same. That has been said before. They look through that telescope and claim to see what everybody else saw without you thinking themselves.

Then he says that the second kind of truth, truth that hasn’t been discovered yet is different. At first it looks too bizarre to be true, and maybe as dangerous as fire. If you’re not clever, it may destroy you. It thus takes cleverness and courage to deal with such new truth. Benno Mueller-Hill goes on to say, you have to get older, perhaps as old as I am, to see that self-deception plays an astonishing role in science. In spite of all of scientists worship of truth.

So I want to say something right now. I think we all need to hear. We can’t use the word research based practices and AI in the same sentence. Yet to to. We are so eager to find research based best practices that non peer reviewed excerpts of articles are going viral before they’re even published. Like the now famous Your Brain on ChatGPT article that claimed that students who use AI show less brain activity. But recently on LinkedIn, Cambridge researcher Doctor Philippa Hardman made a great point.

She said that if you look at other research, it’s not whether students use AI. It’s how the research she was looking at showed that scattered surface level AI use is worse than no AI at all, while strategic offloading paired with higher order thinking can deepen learning, not drain it. The better question we have got to start asking is not is AI making kids dumber? It is. Are we teaching kids the difference between AI usage that makes us smarter and better able to do our jobs?

An AI that causes us to make mistakes in our brains. To stop engaging with what we’re trying to learn. This is what This is why I find people who poke holes in arguments to be so useful. I like to follow people. Maybe even that some call curmudgeons who like tapioca pudding and eat dinner at 445. I may have just described myself, so we’ll go on to a philosopher at Wake Forest that has been asking a different question. What does honesty even mean now on both sides of the desk?

Announcer (31:44): Cool Cat Teacher Talk with award winning teacher Vicki Davis.

Vicki Davis (31:48): Today we’re talking with Doctor Christian Miller, the A.C. Reid Professor of philosophy at Wake Forest University. For over a decade, he’s led some of the largest research projects on honesty and moral character ever funded, including the Honesty Project and The Character Project. His new book is called The Honesty Crisis and it will be out May 19th, 2026. We’re going to talk about something that every teacher and administrator in every college professor is feeling right now.

What honesty means in the age of AI and how it’s under threat in our classrooms from both sides of the desk. So, Christian, thanks for coming on the show.

Dr. Christian Miller (32:30): Great to be with you. I really look forward to our conversation, even if it’s maybe not the most positive thing in the world. It’s an important conversation to have.

Vicki Davis (32:36): It’s the one we’re having in teachers lounges and at the lunch table. What do we do about honesty and where has it gone? Especially in the age of AI? And some people just say, oh, it’s always been this bad as a teacher has been teaching 24 years, I’m not convinced. So what do you think.

Dr. Christian Miller (32:52): On the historical question? I’m not convinced either. I mean, student cheating has been around forever. That’s nothing new. That’s nothing we haven’t heard before. I did think, though, that it ramped up when we shipped it to the internet. So when the internet came along, that was first what I call an honesty crisis, where honesty becomes more appealing and dishonesty is hard to detect. So it’s easier to get away with dishonesty when you have the internet.

That was, I think, the first big moments. You could go online, you could find material, you could put that material into your paper or find answers to problem sets and really cut corners that way. But they didn’t end there, of course. And so with the dawn of AI, I think that’s now the second honesty crisis we’re facing in education. It’s been just progression of more and more opportunity to cheat, as well as difficulty in being caught cheating. That’s a bad combination when it’s easier to cheat and it’s more difficult to get caught cheating.

And that’s a recipe for for dishonesty.

Vicki Davis (33:46): So what does your research say about honesty in this topic? I mean, do people want to be honest? Is it as rampant as we think it is? According to.

Dr. Christian Miller (33:54): Research in general, I think people do want to be honest, but there are limits to going to go so far. So the research I’ve consulted suggests that in general we believe that things like cheating are wrong. We also care about how we’re perceived by others. We want to be perceived by others as honest people. We even want to think of ourselves as honest people. So those are all good things. Those are going to keep dishonesty in check. At the same time, we do have a desire to cheat if we think we can get away with it and it’ll benefit ourselves.

Our character, I think here is complicated. It’s multifaceted. It’s it’s not like simplistic. We’re good, we’re bad, we’re honest. We’re dishonest. I think we’re a mixed bag. As the opportunity to cheat gets easier and the rewards get greater, the good side of our character can erode. So that belief that cheating is wrong will have limits. That desire to think of ourselves as good people in eyes limits. So what is happening with the AI situation is whereas before there was always this threat, maybe I’ll get caught.

There was this idea, well, still going to take me some work to go on the internet and I’m going to have to find this stuff, and I’m going to have to change it and adapt it for my purposes. Now, it’s so easy that the lot of the traditional safeguards aren’t as effective as they once were. So all I have to do is go to this website, give the prompt and I can get in a span of five seconds or 10s my assignment completed for me. And not just just badly or C quality or B quality.

I can get a quality work done for me in 10s, whereas normally that would take me an hour or two. That’s a really hard temptation to resist. Once you know it’s there, it’s available to you. It’s a really hard temptation to resist.

Vicki Davis (35:43): So you’ve also said something that’s caught my attention that in education, the dishonesty problem isn’t only about students, but it’s also about teachers using AI to give feedback. Now, I will admit that I use AI to support me, so I’ll put in my rubric and I’ll grade and I’ll go in AI and I’ll say, hey, this is the points they have off, you know, right this up so that I can paste it in. And I’m kind of, I guess using it like a word calculator. But what’s your perspective on AI to give feedback?

Is it having AI quote grade something I personally think we should never have teachers do? Or what’s your take on that side of the desk?

Dr. Christian Miller (36:22): I think it’s so easy to jump into the student side of it and just focus on that. But I think there’s lots of other aspects to this whole idea. I mean, we could even have a discussion of whether AI itself is honest or not. There’s a question of what about in the teacher’s research if they’re doing research or just newsletter publication, communication with parents, all that kind of side of thing if they’re using AI. But here you’re focusing specifically on teachers grading or teachers assessing student work.

Now, I think it’s all a question of whether there’s transparency or not. So let’s start with a student case. But then where you took us to the teacher case, I have no problem. From an honesty perspective, if a student is transparent that they used AI because now we’re out of the there may be maybe other problems, maybe they weren’t supposed to. Maybe they’re violating the rules. But from an honesty perspective, at least they’re forthcoming. They’re transparent about what they did.

If they don’t do that, if they present their work as if it were their own with no disclosure when it wasn’t, then that actually meets the definition of plagiarism. So they’re using another source to produce the work and they’re taking credit for that work themselves. So that’s just a kind of standard example of plagiarism, whether that’s plagiarism from the internet, whether that’s plagiarism from another student’s paper, whether that’s plagiarism from AI, it still counts as plagiarism, and plagiarism is dishonest.

Well, now we shift to the teacher case. If the teacher says, look, I’m going to assess your work using the help of these tools. Some of my communication with you is going to be generated by AI to help save some time of writing. These emails or whatnot might be other ethical questions that arise, but there’s no in my mind question about honesty. Here’s where I think it gets quite tricky. So the student gets the paper back on the paper are some, you know, typed comments.

Or at the end of the paper there are some type comments or somewhere there’s this feedback. The feedback is AI generated. The student isn’t told that it looks like it’s coming from the teacher, and that’s what any reasonable person would have assumed. It’s not right. They’re actually being dishonest towards the student. That’s if that were to be discovered, that would be a violation of trust. And I think a role of the trust in the classroom, which is very important to have.

Vicki Davis (38:42): Our policy at our school, is that some AI feedback tools are allowed, but with complete oversight by the teacher, complete transparency to your students. So if I have a massive project and and I have AI quote looking at it and supporting me, I always say students review this, this, this is the grade that I’ve given you. But I did use AI in portions of assessing this for you. So if you have any questions or issues, I want you to look this over and bring it back to me.

Is that how we should handle it, or is should we just not even use it at all.

Dr. Christian Miller (39:19): I’m hesitant to tell anyone what to do here. I think that’s great. As far as an honesty perspective, I have no objections at all. You’re being very, very clear. There’s a larger question of role modeling. So in a lot of my work, especially earlier work, I thought about just virtue and character in general, beyond honesty. How should we think about a good person? How should we become better? And one really important idea is role modeling good character in front of other people.

If you want to try and inspire them to be good themselves. So the point here is transparency. That’s great. No problem with honesty. Now, if you’re saying that to the students, they’re going to naturally have the thoughts. Well if you’re using that why can’t I it.

Vicki Davis (40:03): And of course in my classroom they can. But yeah yeah.

Dr. Christian Miller (40:05): So not but some many of them wouldn’t that wouldn’t be allowed. So if it’s perceived as in any way a double standard or hypocritical like you’re not letting us use it, but you’re using it even though you’re telling us why, why not, why not let us use it to that’s that’s going to create problems. So instead, if the idea is I’m going to role model, look, I know these tools available. I’m not going to use them. I’m going to instead pour myself into grading this paper using my own knowledge and experience and skills.

And it might be not as reliable. So I want you to come to me if I made a mistake or whatnot. But at least you know that this is entirely me speaking and I’m not drawing on anything else. That’s that’s another idea. I’m not saying it’s better or worse. I’m writing to tell someone who’s working, you know, tremendous hours in, in K through 12 that they’re doing something wrong.

Vicki Davis (40:57): When we have a student bring something to us. I mean, all of us teachers have a different way of handling this situation. I teach computer science, so we are allowed an AP, CSP to have AI work with a student. I require that they have to be able to explain every single line of code. But say my friend Dawne across the hall, a student brings her paper that is obviously most likely AI written. And we know AI detectors don’t work, which is a big part of the problem.

Is there any research on the best way to handle that possible dishonesty issue?

Dr. Christian Miller (41:32): I have a couple of things to say here that are going to be controversial, and I’ll just flag that. And, you know, listeners, please feel free to to take it for what it’s worth, on the detector front, I would have agreed with you maybe even a year ago or two years ago. And I know you’ve had passed guests talk about this too recently. There seems to be a bit of a change, a pivot. So I’m seeing places like Pangrim, which are reporting stunning reliability and detection.

So almost no false positives. You know, 1 in 10,000 kind of rates of error with false positives. So I am revising my thinking a bit about whether to go to these detectors or not. Even if that’s true, let’s just let’s just assume that there is this great detector out there now. Still, I think we want to have some valid concerns about using it. You know, one is it’s not 100% reliable, even if it’s 99.99% reliable, it’s not 100%. So there’s questions of whether it’s going to meet the threshold for punishment of the students, especially in my case at a university.

We have an actual committee where you would have to go through kind of trial process, and so would we meet the evidentiary bar there. I mean, there’s also larger questions about trust. And, you know, if this is something that’s known to students are going to have to run their papers through that. What kind of classroom environment does it create? Does it break down trust in the classroom and create an atmosphere of fear? Those are all valid points and concerns.

So having said that, I’m of two minds about your original question. So one part of my mind says, okay, a professor, you know, a professor has brought a paper to me that they’re suspicious about. One part of my mind says, let’s use these, these detectors and see what they tell us, what my mind says. Let’s enter a conversation with a student. So I’ve done this before. I’ll say, look, you know, student comes into my office, tell me, tell me a bit about this paper.

Tell me, how did you write it? Walk me through the steps. Can you explain this idea? And this wasn’t something in the class material, you know. Where did this come from? How did you learn about this? At some point, often they will say they admit I use some other source. If they do that, then I’m tend to be quite lenient. I will not take them to any kind of hearing. I’m not going to get other people involved. I’ll say something along the lines. Look, there has to be some kind of consequence for this, but I’m not going to fail you for the class.

I’m not going to sack you for the entire grade. Your grade will take a hit. But let’s learn from this. Rewrite the paper and give you a second chance.

Vicki Davis (44:05): In the detection, you know, my students are pretty open and honest. We had conversations about humanizers, right? Right. Students will whatever AI tool of their choices generate or use it. And then they’ll pull it into a quote, humanize, which supposedly makes it undetectable. And then they’ll run it through whatever AI detection tool they know people are using it. So while a detection service may have false positives, I guess my question is always what you can’t test, which is the false negatives, right?

Like that’s the big unknown here because all it takes is 1 or 2 get away with it. Telling all their peers they’re getting away with it. To discredit any AI detection tool we could try to use. And we want to make this easy. It’s just doesn’t feel like an easy solution and it is a crisis. I agree with that word.

Dr. Christian Miller (44:57): Yeah. So what you said is all very apt and appropriate. So my current thinking is in agreement with you. The detector is not completely reliable. And so I think I echo a lot of people in the humanities at the college level. And I suspect it’s not just at the college level where we think this is a lost cause, where we are at the point where we’re going to take drastic measures to deal with the situation, taking the opportunity of AI out of the hands of our students.

And I will say up front, I know this isn’t going to be a one size fits all for everyone. So what do you do in coding cases? What do you do about problem sets? You know, in the sciences it’s hard, but in humanities I think I speak for, you know, 90 plus percent of my colleagues, not just in my university, but across the country where we don’t think the traditional paper, here’s the paper assignment right at home and then turn it in that has any AI proofing ability anymore.

So we’re just taking that out of the out of their hands and going with either Bluebook, traditional tests, oral exams or some kind of in-class writing scaffolding assignments. This is nothing new to your listeners. I’m sure there’s nothing new to you, but this is one where it is such a crisis that I don’t have a lot of hope that there’s going to be any remedy to help us address it.

Vicki Davis (46:16): Well, just from our perspective as an educator is that we have to be honest with ourselves about the state that AI has put us in with education. Because if we can’t discuss these honestly and say, hey, I’m not buying the line of the tech companies that AI is going to fix everything. AI has broken a lot of things and a lot of us K-12 have moved to oral conversations. My AP class, they have to be able to defend every line of code verbally on the spot.

But then part of my struggle with the handwritten papers is that when you have a student who is dysgraphic and dyslexic, as I do them all family and struggles with written expression, those students literally and legally need technology to be able to help them to truly express themselves in the world. And then there’s the piece of entry level jobs are going away. In lieu of jobs where people can manage AI, they have to be able to manage AI agents where we’re heading and know how to do that type of work.

So we really are in a conundrum, aren’t we?

Dr. Christian Miller (47:32): We are. Yeah. So I have a similar situation, my family. So I’m very cognizant of that. You could make the quick points, but it’s not completely satisfactory that there’s one thing of using a computer and it’s one thing and it’s a different thing of using a computer with internet access. We have we have ways of shutting off or blocking internet access. So at least they have the typing tools and the spell check and whatnot without the option of going somewhere else for help, but that’s not complete.

A completely satisfactory point. Agree with you also on well, in my level of college here, doing a disservice if we’re not providing an opportunity to the students to make use of these tools, because then they’ll be at a disadvantage when they get out in the job world. I don’t know about that. I mean, I think they have plenty of opportunity to use this on their own, and they’re using it all the time anyway. And at least in my world of the humanities, I don’t see I mean, maybe it really varies by field, so maybe I should qualify what I say.

My business students would be different, or my science students would be different. Maybe have to go field by field. I’m not so sure that my humanities students are. I had a disadvantage if I say, look, I don’t want you to use this stuff for my my philosophy papers, I think they’re going to have plenty of opportunity to use it for other things. Maybe I’m here being naive, naively optimistic, but I don’t want to think that I’m disadvantaging them by discouraging them from using it.

Vicki Davis (48:53): Are there any steps that teachers who are listening to you today can make to make their classrooms more honest? Places and spaces? What types of words should be using and how should we be modeling in ourselves?

Dr. Christian Miller (49:07): I would say it starts with how you as a teacher are thinking about honesty. Are you being honest yourself with your students in terms of like we’ve talked about with AI use, being transparent when you’re using it, when you’re not? Or are you keeping that in reserve as something you’re not disclosing? It’s also a matter of, are you honest with yourself? A failure of honesty that we overlook is we can be self deceptive. We can deceive ourselves and not tell ourselves the truth about places where we don’t want to confront something and that bothers us, so we deceive ourselves.

But more concretely, I am a big proponent of the idea that role modeling matters, that we can inspire our students and demonstrate by example. So I think it’s very important to be upfront from day one and expect the students to live by the same standards of honesty that you’re expected to live by, to hold everyone to the same standards here. I really like honor codes. I’m a big proponent of them, I think even though they may not be able to block the AI crisis, they do a lot of good work, and they’ve been documented for 30 years that they do a lot of good work.

So if your school has honor code, I really believe that you, the teacher, should be upfront about your affirmation of it and not just say the students have to affirm it. So in my classes, when they take a Blue Book exam, we all verbally recite the honor Code together. Not just the students. We we all say it out loud. We actually recite it, and I say it too, because I’m showing to them. I’m holding myself to the same standards that you’re being held to.

And when they turn in papers, I make them handwrite the honor code and sign it. So even if they did something fishy on this paper, they still have to take a stand at pledging their integrity and honor in writing that honor code and signing it. That’s taking a stance, and it can be held accountable for that. So I think both sides affirm an honor code, and the students write it out on papers and verbally recite it on exams. It can’t hurt and it can only help.

It won’t fix everything, but it can only help.

Vicki Davis (51:10): So Christian, what is your goal for the book The Honesty Crisis?

Dr. Christian Miller (51:14): The book is larger in scope than just education, so it’s talking about six different areas of society where I see honesty crisis happening and education is only one of them. Others include things like deepfake videos, political misinformation, celebrity, even religion. I talk about things like sermon plagiarism. So I’m I’m worried about honesty crisis popping up all over the place. I guess my goals would be threefold one just appreciate what honesty is in the first place.

Get a better handle on what it is to be an honest person. That’s actually quite interesting and more nuanced and complex than you might have thought. It’s not just a matter of don’t tell lies. It has way more richness to it than that. So greater appreciation, whilst he is greater appreciation of where it’s eroding in society, places. We might not even have paid attention to. AI in the education case is an obvious one. Some are more subtle. Lastly, I kind of feeling and a motivation to do something about it.

Not only now I’m I’m better equipped a more knowledgeable about this, but I’m also inspired to change my life and work towards change in other people’s lives to address these honestly. Crisis. Because honesty is our most treasured virtue. when it erodes, terrible things tend to happen in society, so it’s worth fighting for.

Vicki Davis (52:30): Honesty is worth fighting for. As an educator, I want to believe in the integrity of what’s happening in my classroom. We don’t want to give our lives for something that’s pretend, that’s fake, that’s not real. And we know that people struggle with, you know, what do I believe in in the world anymore? You know, more than ever, kids do need to see teachers who do have honesty and integrity and understanding. All of us are flawed, of course. But as I say to my kids, when you mess up, fess up.

Dr. Christian Miller (53:01): I agree entirely. Yeah, the cover up is often worse than the crime. At least be honest about the wrongdoing.

Vicki Davis (53:06): Yeah, so we’ve been talking with Doctor Christian Miller. The book is The Honesty crisis. As I always say, we don’t play King of the Hill. We make a bigger hill. There’s a lot of room for conversation. And these are the things that we need to be talking about in our staff meetings at the lunch table, as we move forward and work to make the world a better place. Thank you, Doctor Miller, for being part of that conversation.

Announcer (53:28): Cool Cat Teacher Talk with award winning teacher Vicki Davis.

Vicki Davis (53:32): So as we talk about honesty, let’s come back to our opening story by Benno Mueller Hill from 1993. In the Quarterly Review of Biology, about 38 boys in line who looked through a telescope and all claimed to see a planet. But it took the 39th boy to insist he couldn’t see anything before the class discovered that the lens cap was still on the telescope. Now is not the time to have blind acceptance of what everybody else is saying. In fact, I want to make a point that the story I’ve hinged this whole show upon, I wrote down on an index card in 1995 and put it in my quote box.

AI would have never found that article because it is behind a login wall and it’s from 1993. I couldn’t even use AI to fact check my story, but it did find the journal and I was able to log in and read it again myself before I shared this story. AI and the internet has a recency bias. When you’re using AI, it is not searching the wealth of the ages. In fact, there’s a new AI that’s been trained on everything from the 1800s. We have this recency bias, and if we don’t watch out, we’re going to leave out the wisdom of the ages and make the same mistakes others have made.

Let’s look at when radium was discovered. People believed that radioactivity was a secret to long life. So they made this thing called a revigator. This was a big ceramic holder of liquids. You would dispense the liquids and then drink them, and they claimed it would extend your life. It was lined with radioactive clay. Some people put water in it and other things. But if you were unlucky enough to put juice in your revigator just a few years later, you would be dying a painful death from cancer.

It took time for the truth to be known, to try out AI for lots of tasks, and it might work for a third of them. The other day I was shopping for a 4K webcam to upgrade the quality of the television version of this show, and Claude recommended a certain webcam. Highly rated, and once I got it installed it, I found out that I couldn’t shoot in 4K in Riverside, the recording platform I use. I made a mistake. Now granted it was a small mistake, but still think about all of the decisions that people are making using AI that are going to cost them dearly.

Every time I see someone say they are breaking up with their girlfriend of a year or two because of advice from ChatGPT, okay, ChatGPT doesn’t date, it doesn’t get married, it isn’t human. And if you go back and say it was wrong, it will go, I was wrong. Sorry. When you’re having terrible consequences for implementing the decisions it so casually gives you, when we blindly follow those looking through the telescope and claiming some sort of amazing insight of AI and how we’re going to live longer and happier lives and how we should use it.

And when we look through the telescope and we don’t see that and we don’t say a word, we’re being complicit in more people making blind mistakes. If we’ve learned anything today, it is that we have a very powerful technology, but we don’t know how to use it yet. There is a specific way that we have to operate in times of fast change. We need to be experimenting. We need to be testing. We need to keep what works and stop using what doesn’t. Don’t play King or Queen of the hill.

We need to make a bigger hill and include more people in the conversations about what AI can do to improve learning, and not kill the hopes and dreams of kids who aren’t mature enough to know yet that dishonesty in learning will only cheat themselves of their future. My thanks to Justin Reich at MIT is limited series. The Homework Machine is wherever you get podcasts, and to Doctor Christian Miller at Wake Forest. His book, The Honesty Crisis is out May 19th, 2026.

Wherever you get books, links are in the show notes at coolcatteacher.com/honestai. I’m Vicki Davis and you’ve been listening to Cool Cat Teacher Talk. See you later, educator.

Announcer (57:55): Cool Cat Teacher Talk with award winning teacher Vicki Davis.