On the Podcast: AI in the Writing Classroom, Part 1

I think sometimes the right book comes at the right time, and I just had that feel in reading your book, how needed it is. There's such a titanic shift in education right now, and I work with teachers who embrace AI. I work with teachers who are afraid of AI, so I've been looking forward to this conversation. One of the things that you talk about in your book that I'm really... Found fascinating is that you want to position AI as sort of a writing partner.

Absolutely. I think when AI was first released, my biggest fear was that students would use it to cheat or that they would become over-reliant on it. And I instantly, kind of naively, thought that being a workshop style teacher would be the savior, that none of my kids are going to use or want to use AI because I write alongside them and we workshop together. But it was maybe eight weeks after AI's release when I caught a student using ChatGPT in class on an assignment that we were writing together.

And the conversation I had with the student really opened up my eyes to the fact that AI can be a writing partner. Specifically, he said that this rough draft that AI generated about why he plays football for him was really similar to the mentor poems that we had done in the unit, and it kind of helped him get started and get through the writer's block of writing. And I thought that's a really interesting way to look at AI. And I know that I can't spend the rest of my teaching being a plagiarism investigator. It's not why I became a teacher. So if I can harness it to help my students with their writing so that they can author some really compelling pieces, why wouldn't I jump on board?

So we've given a couple of presentations to our own school, and in those presentations we started with this similar concept of you have this choice of avoiding AI or we can embrace AI. And one of the stories that I told was going back to the time of social media, when I first started teaching, it was really kind of hitting big in schools. So there were Facebook projects where kids would break down the idea of Katniss Everdeen through a Facebook page or something like that. And we were really on it at that point with social media.

But as the different forms of social media appeared, we fell off of that. And then social media really kind of claimed our student population in a way that I don't think that we were happy with. I don't think that anybody was really happy with the aftermath of what social media did to our children. And I would say that AI has a potential to end up in that place if we don't pay attention to it, if we don't teach it in ways that it could be used effectively as opposed to ignoring it. I would fear AI more if we ignored it, like we ignored social media in terms of an educational context than I would if we engaged with it and taught students how to use it as a tool, which is just another component of digital literacy.

Not only how to use it properly, but just how it's going to make you a better writer. Right?

And a better thinker.

And that cow has left the barn, it's not going to go back in.

Right. I would add too to that, because AI trains itself on what we input and who is inputting the questions or the prompts. If we leave out a certain population, maybe an underserved population, then AI is going to keep getting trained on the biases and hallucinations it already has. In order to train that out of it, we need to make sure that everyone has access to it.

It raises an interesting question too, because the feedback that you're going to get is only as good as the questions that you're going to ask it. And I think one of the things I really appreciated about your book was teaching kids to ask the right kind of questions.

Exactly. It's important. It's that concept that if you type in something that's better for a search engine like Google, you're going to get a really vague response. And that ability to re-prompt an AI and say, no, try again this time, I want three examples or I want your sources is a really powerful way to harness and leverage that tool.

AI is a flipping of the switch I've noticed as well. Kristina's been the master of flipping the switch early and also of prompting. She's the one that I go to for prompting. But that flipping of the switch is important too for teachers to think about if you're not using it at all, there needs to be a moment where you take a pause before you go in the direction that you're traditionally used to going and ask yourself, how could AI help me with this? She's often referred to it as her intern. Well, let me ask my intern. It took me a while to think of AI as an intern, as a resource that I could go to or should go to for an answer or for an avenue or for my next idea.

That's really interesting because I think a lot of teachers just see, oh no, here comes plagiarism.

Or over-reliance.

Right, that it's going to do the thinking for me. Even long before AI, if you ask a kid to analyze a novel really for an underdeveloped essay, it was always just an extraction exercise. I'm just going to go find two or three reasons. I'm going to pull them out a book, drop on the page. The separator was the commentary. And so I'm wondering how AI helps you think deeper about the things you want kids to think about.

Yeah, I just got done teaching Macbeth and I had my students track quotes the entire time, like I have always done traditionally as we're reading. And then they get to pick their own prompt, whatever they want to write about. So I had a student who wanted to write about the liquid imagery. I had a student who wanted to write about the witches. We had a long conversation about the dagger scene. Was that guilt hallucinating it or did the witches cause it? And with generative AI and teaching them how to prompt it, you can have it take on a perspective or a role. So they chose a prompt engineered response, which I put into my chatGPT and they got to pick the role. So they got to say, act like the witches from William Shakespeare's play Macbeth and answer these four questions. Did you mean for Macbeth to kill Duncan?

Did you cause the dagger scene or Banquo's ghost, or was it just the guilt? And then it would respond in rhyme like the witches might speak, or if they prompted it like Macbeth or they asked who the third murderer was, was that a mistake on your part, William Shakespeare or not? And then they get to pull and they're excited, shockingly, to write an essay on Macbeth because they get to pull these cool things and they're using it like another voice in the conversation and they can quote from it in their essay, which they're very excited about.

So they pull that and then they comment on that?

Exactly.

They analyze that?

They analyze ChatGPT's.

They agree, they disagree?

Yeah. I have guiding questions for that. What did it get wrong? Are you smarter than it? Because in some instances it still gets things wrong, or would you have picked a different quote or a better quote? And then they get to write about that. And I grade more their thinking and their ability to embed quotes than I do their topic sentences or anything like that. I just want to see their thinking.

I'd like to highlight the fact that you are encouraging them to question the output as opposed to just accept it at its face value. So packed into that assignment is questioning the material that you're pulling from the internet, in this case from AI. And that's really important. That's one of the skills that we want them to take from their experience, whether they were just on the internet or reading a book or using AI in that particular case. I would just say too, Kristina is talking about an experience for students in the classrooms where teachers, I think, would be more willing to go to a one-on-one level with students and allow them to prompt AI.

For someone who's less interested or ready to approach AI in that level, you can go up to a kind of full classroom perspective. And I've prompted it to give me five comments on, since we're talking about Macbeth, the imagery in Macbeth, what's the most important imagery? You're asking AI app, you print it, you bring it into the class. We're going to have our class discussion on these comments today. What do you agree with? What do you disagree with? What does it get? What does it miss? And there's a little bit more control in that kind of handout compared to the student to student. And that might be an entry point for some teachers.

Are they doing some sort of reflection when the essay is over?

Yes.

We have that as a component of our rubrics. We do reflective rubrics oftentimes, which has got maybe the standards or the craft that we are studying in the left-hand column that's just describing it. They will implement or import their own sections of their writing in there and reflect on that. And at the end of it, we ask, did you use AI? And if you used AI, how did it help you? Or how did it not help you? And we've had all different variations of descriptions from students. Some of them just don't use it. Some of them use it quite extensively, and we have really wonderful narratives about the trade-off that they had with it. Some of the examples we've given earlier about their beliefs about what AI did for them and what it's not allowed to do for them.

Thanks so much for tuning in today. You can visit blog.heinemann.com to learn more about Dennis and Kristina's new book and purchase the book. You can also read a full transcript while you're there. And this is just the first episode in a three-part series, so don't forget to tune in for the next two episodes.