Sponsor MessageBecome a KQED sponsor
upper waypoint

‘Not Even AI Can Save Me’: Students and Teachers on ChatGPT in the Classroom

Save ArticleSave Article
Failed to save article

Please try again

Isabel Toscano's classroom at Oakland High School in Oakland on Oct. 9, 2024. (Martin do Nascimento/KQED)

Artificial intelligence inside California schools feels a lot like building the plane as it’s flying. Students who are familiar with tools like ChatGPT are using it faster than teachers and administrators can regulate it. Jesse Dukes, co-host of the Homework Machine podcast, talked with dozens of students and teachers across the country about their thoughts on AI in the classroom.


Some members of the KQED podcast team are represented by The Screen Actors Guild, American Federation of Television and Radio Artists. San Francisco Northern California Local.

Links:

This is a computer-generated transcript. While our team has reviewed it, there may be errors.

Sponsored

Ericka Cruz Guevarra [00:01:28] It does feel like AI is sort of everywhere now, including in classrooms. Is that pretty accurate? Is it safe to say that AI is just everywhere?

Jesse Dukes [00:01:44] It depends what you mean by AI, right? I mean, if we’re talking about generative AI, which is, you know, chat GPT in this generation of AI, it’s certainly made its way into schools. When you’re talking about what do students use, either in ways that their teachers approve of or in ways their teachers disapprove of, it’s gonna be like chat GPT, I would say like 90% of the time.

Ericka Cruz Guevarra [00:02:15] Well, I mean, you’re in the thick of it in a sense because you did a bunch of interviews with both teachers and students for this story. And I know you met a student named Emilia. Can you tell me a little bit about Emila? Who is she and where is she from?

Jesse Dukes [00:02:32] So I talked to a student named Emilia who asked to stay anonymous. She’s from somewhere in Northern California and she was a senior last school year when I spoke with her.

Emilia [00:02:48] My schedule is from 8.30 to 5 being at school and then from 5 to 6 I have soccer. So it’s just like I love being at school. I don’t like being at my house.

Jesse Dukes [00:03:01] She says that she likes to learn too. And she actually, when I was asking her, do you see students using AI to do their schoolwork or do their homework, she said she did and expressed some disapproval of that.

Emilia [00:03:17] You’re not learning. You’re just learning how to copy and paste. Why would schools be invented then? Or like, why would we come to school? Or like it just like, it just applied.

Jesse Dukes [00:03:30] And she also described herself as somebody who really likes writing as well.

Ericka Cruz Guevarra [00:03:34] Okay, so she sounds a little bit like a rule follower.

Jesse Dukes [00:03:37] Yeah, usually.

Ericka Cruz Guevarra [00:03:39] That is until she it seems hit a wall with one of her classes. What happened?

Jesse Dukes [00:03:49] There was a teacher who I believe was new to the school, an English teacher, so somebody who didn’t know her, and Emilia said that when she was turning in writing assignments, the teacher just didn’t seem to like her writing.

Emilia [00:04:03] All the essays that we have been doing. She has given me either a C. Or an F.

Jesse Dukes [00:04:10] And Emilia found that confusing because, you know, as I mentioned, she thought of herself as a good writer.

Emilia [00:04:15] I actually took a college class with a professor in UC Berkeley and I never had these type of like, your writing sucks, uni, so much and so much. I always had A’s in that class and the professor even liked so many of my writings. I was just like, how am I getting A’s, in my college classes, and with you, I’m getting F’s and C’s.

Jesse Dukes [00:04:42] She’s in the class, she’s frustrated, and she keeps getting grades she doesn’t like on writing assignments and one time there’s an in-class extra credit assignment. So it’s sort of low stakes, right? You’re not gonna get a bad grade if you don’t do a good job. And she thinks, I don’t know, I wonder what will happen if I get ChatGPT to do this for me.

Emilia [00:05:04] Yeah, I mean, I could just write it myself, but I was just so tired, I was like, I’m just AI-ing.

Jesse Dukes [00:05:15] So she gave it a try. And then apparently her teacher was able to detect that she had used ChatGPT. And so then Emilia got this message, I believe it was in Google Chat, that essentially said, I don’t think this is your work and you’re not gonna get any extra credit. And if you wanna dispute that, you can come talk to me. But Amelia didn’t dispute that because she knew that the teacher was correct.

Emilia [00:05:42] That was like my whole jaw dropped and I was like not even AI can save me.

Jesse Dukes [00:05:48] She had this kind of pit of the stomach moment, as many of us would when you get caught doing something embarrassing or against the rules and maybe something that in Amelia’s case goes against her own values.

Ericka Cruz Guevarra [00:06:00] Totally, totally. I’m just thinking like how I was as a student, that’s like my worst nightmare.

Jesse Dukes [00:06:05] Yeah, no, I get a little bit of a shiver just thinking about it.

Ericka Cruz Guevarra [00:06:09] I think I can easily imagine how a student might use ChatGPT for like a social science class or an English class, but how can it be used for maybe like a math or science class?

Jesse Dukes [00:06:21] It’s pretty good at math. You can just transcribe the math problem into ChatGPT, but there’s also something called Photomath. Have you seen Photometh? Which is an app that integrates the phone’s camera and you can just hold Photomoth above the math problem, even if it’s hand written and it will solve it for you and output an answer. And it even works with word problems.

Ericka Cruz Guevarra [00:06:46] I gotta say, as someone who is not good at math, I might have been tempted by that as a student.

Jesse Dukes [00:06:53] It’s very tempting. So Justin, my co-host, tells a story about how he was helping his 14-year-old daughter with her math homework, and they were both stuck. And the daughter just whipped out her phone and put it into photo math, and they used it to figure out how to solve that problem. Now, according to Justin, then they put the phone away and tried to do the rest of the problems without using photo math. And actually, this use case I just described to you, I think this is a classic gray area, if you’re stuck in your homework. Many teachers say you can use resources. Can you use photo math as a resource or ChatGPT as a research to figure out how to solve the math problem, particularly if you then put it away and solve the rest of them using your own brain? It’s a little bit unclear right now.

Ericka Cruz Guevarra [00:07:36] I want to talk more, Jesse, about some of the conversations you had with educators. You spoke to dozens of educators about how they’re feeling about AI. And I feel like whenever I hear conversations about AI in classrooms, the main concern is always, as we’ve been talking about, this idea of cheating. But what are some of other concerns that you heard from educators about AI right now?

Jesse Dukes [00:08:03] Cheating is probably the biggest one. One of the major concerns around cheating is, you know, there’s this sort of ethical concern, but there’s also the concern that if the student is using AI to do their homework, they’re bypassing the learning that the teacher is trying to get them engaged in. They’re bypassed the thinking. Other concerns, will what we’re trying to do… To keep students from cheating end up having its own harm? Are we going to punish students unfairly or harshly who are suspected of AI cheating? Is that discipline going to be unfair or inequitable in some ways? There are concerns among teachers that teachers might be replaced by AI. I don’t think that is imminent, but there are a number of pilot programs of AI-powered tutors right now. And then I think there’s a concern, this sort of existential concern about, are we teaching students what they are going to need to be successful in an uncertain future that may very well have more AI in it? It’s an age old question. Are we teaching the right things in schools? And AI has exacerbated that question.

Ericka Cruz Guevarra [00:09:21] And you mentioned earlier this idea of disciplining students and how to discipline students for using AI, which feels tied to what policies schools have around AI and whether those even exist. I mean, are there policies around AI for schools in California?

Jesse Dukes [00:09:41] It’s really school by school, district by district. We did a survey with the RAND Corporation that found that in early 2024, so about a year ago, only about one in four teachers said their schools had an AI policy or had gotten any AI guidance. Both the teachers and the students are saying it can be a little bit confusing what counts as a responsible use of AI by a student and what counts is cheating. If we did that survey again now, we’re now in the fall of 2025, I suspect that more schools do have policies written. More schools have probably had these conversations about how do we want our students to use AI responsibly? What kind of guidance do we wanna give our students? What do we consider cheating? But it’s taken a while.

Sara Falls [00:10:37] AI keeps changing and keeps getting better and keeps more and more tempting and more and accessible.

Jesse Dukes [00:10:46] Sara Falls is a veteran teacher, English teacher, who teaches at Abraham Lincoln High School in San Francisco. She’s been teaching there at least 20 years.

Sara Falls [00:10:57] I trust a lot of my methodology, I trust a lot my processes. It’s not just about writing, it’s about thinking and I’m really good at helping kind of unpack and help students to think critically.

Jesse Dukes [00:11:11] For her, AI has been kind of horrifying, seeing it come into school, seeing it coming to the classroom.

Sara Falls [00:11:17] I’m starting to worry that the career I’ve built that’s really about deepening student critical thinking in order to create something meaningful is going to just go out the window. That people are going to say, my ability to think critically about this and write about it well does not matter. That, for me, is the biggest issue.

Ericka Cruz Guevarra [00:11:39] Yeah, I mean, how did she see her students maybe using AI in the classroom and how did she get her school to address it?

Jesse Dukes [00:11:47] She really wanted her school to adopt a kind of zero tolerance policy when it came to AI. And around that time, she reached out to the school district for guidance, and she said due to personnel issues and staffing issues, the person at the district level who would usually, whose job it would be to think about these things, that person’s job had been eliminated and not replaced.

Sara Falls [00:12:10] I would like to feel like SFUSD leaders recognize that this is a concern, and especially, again, as teachers of writing, what are we going to do about this? Where does it become a really useful tool versus it’s getting in the way of students’ ability to create and draft and revise meaningfully on their own?

Jesse Dukes [00:12:32] The district did have some guidance about AI, but it was pretty basic. So she thought we were just going to have to come up with our own guidance for the students. And so she started a process to write a responsible AI policy for her school, uh, roped in other members of the English department. They wrote a draft and they shared it with the rest of the school. And then my understanding is they went through some revision processes and then she presented it to her school’s leadership over the summer.

Sara Falls [00:13:06] The first to do is some deterrence, trying to explain what the concerns are. It’s detrimental to the environment. It uses copyrighted works and is creative theft. If using Grammarly, be an active editor, don’t necessarily accept all the suggestions, their style isn’t always better. Google Translate, use it to look up words, but not, don’t write in one language and then run the entire text through Google Translate. Mostly what I wanted was to feel like We’re all sort of discussing this and deciding that this is a district-wide concern and let’s figure out how we’re going to move forward with it.

Ericka Cruz Guevarra [00:13:49] We’ll be right back with more from Jesse Duke’s co-host of the Homework Machine podcast right after this. Stay with us.

Ericka Cruz Guevarra [00:14:05] So Sara is the sort of AI skeptic, it sounds like, but I know you did talk with some teachers who find it pretty helpful, actually. What did you hear from those teachers about positive uses of AI?

Jesse Dukes [00:14:22] Well, for one thing, you know, teachers are sort of even handed. So even the most skeptical teacher, you know, because they’re sort of like journalists in this way. You’re like, well, can you think of anything good about AI? They’ll usually come up with something, right? They’re like I used it to sort my students into their field trip groups. So there are these organizational tasks. I’ve had a number of teachers tell me that they use it to write their self-evaluations, which, gosh, there’s a lot to say about that. There are some teachers who describe uses of AI to really support learning in interesting ways.

Eric Timmons [00:15:00] I was reading about OpenAI and ChatGPT. And so I started playing with it. And like everybody else, I was like, well, this is pretty incredible.

Jesse Dukes [00:15:10] So one of those enthusiasts was Eric Timmons, who is a film and English teacher at Santa Ana High School. He teaches film. His students are majority Latino, majority low income. And he talks about using AI to help him come up with kind of like teaching ideas and curriculum ideas. And I should say, he’s a veteran teacher.

Eric Timmons [00:15:36] The part of our job that is sometimes cumbersome of like putting the curriculum together and all these plans and stuff like that, it makes it really fast.

Jesse Dukes [00:15:43] He was giving his students an article to read about gentrification in Santa Ana. And, you know, this was an article he found. It wasn’t something the school had given him. So there were no discussion questions. There were no bridge activities. So he wanted to come up with some kind of in-class discussion framework. I was like, okay, let me try and apply ChatGPT and come up with a structure for this. So he asked ChatGpT to design some discussion questions He’s somebody who’s taken some classes from different curriculum designers. And he’s a fan of something called Project Zero. And they have a particular bridge activity framework that he likes. So he asked ChatGPT to design some discussion questions in accordance with that Project Zero’s bridge activity. And it sped out some questions for him. He kept the ones he liked, he got rid of the ones he didn’t think were gonna be very effective, and that gave him his in-class discussion activity, you know, in like two minutes.

Eric Timmons [00:16:44] As a teacher there’s like sometimes a feeling of like well you know what like I’m not a curriculum designer like should I be going and doing all of this work for free you know um but at the end of the day it’s not going to happen so I have to do it right being able to work with the smartest curriculum designer that’s ever existed and have that at at my disposal is incredible

Jesse Dukes [00:17:07] And I think one thing that, you know, my co-host Justin Reich, who’s a professor of education, would be very quick to point out probably is that there does seem to be a pattern that the teachers who use generative AI effectively often tend to be the veteran teachers who already probably could do the thing they’re asking chat GPT how to do. First of all, they’re able to know what to ask for in very, very specific terms. And they’re also. Able to evaluate what the AI bot gives them and keep the good and dispense with the part of that that’s less useful.

Ericka Cruz Guevarra [00:17:49] I’m wondering how myself and just people who are maybe seeing what’s happening in classrooms around the country and California, how we’re sort of supposed to make sense of all of this hype and this push around AI by our state leaders, by AI tech companies, and how we sort of square that with what you’ve heard from students and teachers and the People actually like living it in these classrooms.

Sponsored

Jesse Dukes [00:18:20] I think we should all be a little bit skeptical of the idea that AI is going to transform and democratize education and do so quickly. And I think that we don’t really know what skills students are going to need in the future, even if there is way more AI in the workplace. And historically, we’ve not been very good at predicting what skills our students are going to need to succeed in a world that is being increasingly redefined and shaped by different technologies. But I think the absolute most important thing that schools need to be doing right now, and I’ve definitely heard this from teachers, is they need to creating space for the kinds of conversations between teachers, between teachers and students, between school leaders about what values they want to center when they are giving their students guidance about how to use AI responsibly. You know, what counts as responsible use of AI? What counts as cheating? There are gonna be people in the schools who disagree about that. And if they can bring students into those conversations, all the better. But too many schools haven’t really convened those conversations. They haven’t given their teachers the chance to discuss AI, responsible use of AI with one another. And what I’m hearing from teachers is that when they’ve had a chance to be exposed to a little bit of information about AI and to talk these things out with their colleagues, with students, with school leaders. At least they feel a lot more comfortable and they feel like they can work towards a framework for how to proceed in this pretty complicated and confusing moment.

lower waypoint
next waypoint