Martin Puchner, the Byron and Anita Wien Professor of English and Comparative Literature at Harvard University, is a prize-winning author, educator, public speaker, and institution-builder in the arts and humanities.

You’ve been playing around with customized AI chatbots, creating spaces for students to chat with famous figures in history (Confucius, Shakespeare, Woolf, and Du Bois). What has that process revealed to you? How are your students reacting? 

It’s been a fascinating journey. While I have extensive experience teaching online, I certainly don’t consider myself a tech expert. I don’t even know how to code. But over the past few years I have been invited to a couple of tech conferences, as a representative of the humanities. At one of those conferences, in January of 2024, I learned it had become possible to create customized chatbots without needing to know how to code. I started to play around with the technology and was amazed how relatively easy and interesting it was. It made me realize that much of what had driven my work—how to make the past speak to the present, how to read ancient texts, and what it meant that the words of long-dead authors were available to us in the present—was relevant for interacting with chatbots.  

The first chatbot I created was based on Socrates in part because I noticed that we interact with generative AI through a chat interface—and “chat” is just another word for dialogue. Many years ago, I had written a history of the philosophical dialogue The Drama of Ideas, so I thought: this interface lends itself to a dialogic thinker such as Socrates. As I was “talking” to Socrates, it occurred to me that this Socrates chatbot was just an extension of what Plato had done over two thousand years ago, when he turned the live interactions with his dead teacher into written dialogues. When we read Plato, we activate this dead figure, Socrates, in our minds, based on Plato’s text. The chatbot does something similar, only it turns writing back into a live, interactive format, precisely the format Socrates favored.  

Students have been reacting to the chatbots with great interest. Many are using them to explore the layers of thought that can be found in these figures. But others are using them to explore more existential questions. Apparently, Nietzsche and Montaigne are especially helpful in offering life advice. 

Do you have a favorite figure to “speak to” in this way?  

I have different conversations with different figures. With Socrates, I talk a lot about the internet because his worries about writing resonate with our fears of fake news and the effects of cell phones. With Nietzsche, I like to provoke him by asking naïve questions, because this makes Nietzsche a bit more direct, not quite as polite as the other chatbots—he tells me when I’m wrong. For example, I asked him whether his famous attack on Wagner was a result of personal jealousy, and he told me in no uncertain terms how wrong I was. With George Eliot, it’s great to talk about realism and literary style. I also built a chatbot in which Scheherazade provides writing advice based on the 1001 Nights. Oh, and I did a bot that impersonates the Creature from Shelley’s Frankenstein. Shelley had helpfully specified the “training data,” i.e., the books the Creature reads, so I was able to use those. 

What were your initial goals when you started working with AI? What can students, and instructors, learn by engaging with artificial intelligence?  

I didn’t have a goal in mind; it was purely exploratory. But I quickly realized that they were great teaching tools, so I made more of them and made them available on my website. The way you create them is to upload a text (should be in public domain) in html and then you add instructions on what the bot is supposed to do with the text. So, in the Socrates example, my instructions—which I arrived at through trial and error—focus on two areas. The first is to create a character who lived at the time of Socrates, who speaks as Socrates, who uses the Platonic text I uploaded as a basis, quotes from it, etc. The second focus is to define dialogic behavior. Normally, chatbots will spit out bullet points or long answers. In contrast, I tried to make them engage in a real dialogue by varying the length of the answers, asking questions of the user, essentially modeling the dialogic form of the Socratic dialogues. 

The biggest takeaway, besides the individual conversations, is that we teachers in the humanities often think that what we’re doing is very different from newly emerging technologies like AI. I used to think that myself. But in reality, AI is an extension of the basic technologies that we use all the time, especially in writing. This shift in mindset has practical consequences in that it might encourage us to work with these new technologies. But it also has more philosophical consequences about what it means to interact with AI. It’s something I find myself mulling over a lot. Questions such as what does it mean to be creative; how does something new emerge; to what extent are we relying on tools.  People have asked these questions before, of course, but AI is now forcing us to confront them in new ways. 

Your own website mentions using “time travel to understand big ideas.” How do you make remote time periods feel fresh and relevant for today’s students, both in playing with GenAI and in other ways? 

For historians like myself, history is interesting in and of itself, but that’s a minority view. For most people, you have to make it relevant, and I think that’s a reasonable demand. The chatbots are one response because dialogue is such an intuitive way of engaging with ideas (which is precisely why Socrates favored dialogue). But there are other ways. In my writing, I use narrative strategies, such as unfolding a historical moment by focusing on a human agent who is making decisions in a particular context, or a resonant object. It also helps to raise urgent thematic questions regarding climate change or migration, and look at how they have played out in cultural history. 

Many of the figures featured in your chatbots appear in your latest book, Culture: The Story of Us, From Cave Art to K-Pop, which you’re expanding into a textbook for introductory humanities courses. How have your virtual conversations informed the way you approach textbook writing? 

The whole chatbot thing is new for me so mostly it’s been the other way around: I’ve used what I have learned from writing, especially The Written World (where I talk about dialogic philosophers) and Culture (where I focus on objects or historical figures) in making my chatbots. But you’re right, the influence should also go the other way. I’ll definitely make my chatbots available for students who use the textbook, and I’m also thinking about creating a chatbot version of the textbook. Perhaps I’ll also ask some of my chatbots, especially those like Nietzsche or Montaigne who developed distinctive styles of writing; or Margaret Cavendish, my latest chatbot, a seventeenth-century philosopher who also wrote speculative fiction. 

What advice do you have for teachers who are considering incorporating generative AI into their humanities courses?  

My guess is that few teachers would object to their students having conversations with philosopher chatbots. My colleagues are more worried, I think, about students using generative AI to write essays, and to some extent I am as well. I’m thinking about this a lot in the context of an online writing course I’m developing. My team and I are working out very carefully when and how to use AI—and when not to. AI should not be used to churn out a first draft. But if used right, AI can be used effectively as a sparring partner, to produce a reverse outline of a messy first draft and even, if prompted carefully, to evaluate a first draft. So, I do think that AI can be used as a writing tutor, as long as students don’t use it to outsource their thinking and writing. 

Interested in contributing to the future of teaching in the humanities? Visit this link to sign up to serve as a paid editorial reviewer or content consultant for Puchner’s forthcoming introduction to humanities textbook. 

MEET THE AUTHOR