Martin Puchner sitting next to a flower bed.

Martin Puchner.

Niles Singer/Harvard Staff Photographer

Arts & Culture

A modern approach to teaching classics

Martin Puchner is using chatbots to bring to life Socrates, Shakespeare, and Thoreau

6 min read

Making education accessible to a wider range of students is the driving force behind many of Martin Puchner’s projects.

It’s why, in the mid-2000s, he took on the daunting task of editing the vast “Norton Anthology of World Literature,” which is used to introduce college students in classrooms around the country to classic texts.

It’s also the reason he now experiments with building customized AI chatbots, allowing students to speak directly with famous figures from history, such as Socrates, Shakespeare, and Thoreau.

“I just think there’s so many barriers to education, that wherever I find an opportunity to lower them, through technology or pedagogical devices like an anthology, I’m game,” said Puchner, the Byron and Anita Wien Professor of Drama and of English and Comparative Literature.

Puchner sat down with the Gazette to talk about the anthology, which published its fifth edition this month with a new feature that provides different translations of text for students to compare — including one done by AI technology. The interview has been edited for length and clarity.

What is it like to edit an anthology that encompasses such a vast range of global literature?

No one is trained in literature on that scale. In retrospect I’m amazed how little emphasis on that big picture is placed in education. But it allows you to see patterns and developments that you’re not able to see if you look at literature, culture, history in chunks of 100 or 200 years. For me it’s just been completely life-transforming. Everyone who first encounters such an anthology, including teachers, will say, “There’s so much in here I didn’t even know existed.” Since this is for American students, it’s a way of learning about the rest of the world. There is some American literature in here too, and some English literature, but the emphasis is on everything else.

“I thought, ‘I wonder what happens if I can use the dialogic form of the chat to access dialogic philosophers like Socrates?’”

Illustrations of famous authors in grid format.

Customized AI chatbots allow you to converse with historical figures, based on their words and ideas.

This anthology includes more material from oral storytelling traditions from Africa and the early Americas, tell me about that.

We started to think more intentionally about how to include oral literature, including Native American literary traditions, African literary traditions. We’d always had texts in the anthology that had been transmitted orally before they were written down (for example, the West African “Epic of Sunjata”), and we always had a cluster of fairy tales and folktales that were written down in the 19th century. But we didn’t really think systematically about it — we tried to do that this time, especially with respect to chronology. An anthology is organized mostly chronologically by when something was written down, but that doesn’t work very well when something has been transmitted for centuries and may be written down as an afterthought or by an anthropologist. We decided, at least in the thematic clusters, to mash up the texts that were written down in a given period with others that were written down later but that had an interesting thematic connection.

Tell me about the new “Translation Lab” feature in this edition.

Translation is a fascinating topic, because that’s what makes world literature flow. No one knows all these languages. That means if you want to do big-picture or cross-cultural reading, you have to do it in translation. Sometimes that’s treated as an embarrassment — scholars in comparative literature can be very snooty about that — but there’s an amazing amount of insight you can get out of good translations. It’s also cool to see how different translators approach that from different angles. Rilke’s “Archaic Torso of Apollo” is a modernist poem, it’s extremely obscure, it’s very difficult, including in the original German. We have a bunch of translations: One of the translations is by the wife and co-founder of Norton, our publisher, another is by myself, one is by Google Translate. I thought that would encourage students to think about what goes into translation. We encourage students to try out their latest translation software and see how these different machine translations handle difficult metaphors.

You have been creating customized AI chatbots that personify historical figures like Socrates, Aristotle, and Confucius. What inspired you to create these?

We interact with AI through dialogue. Thinkers in antiquity, Socrates, Confucius, the Buddha, lived in literate societies, but none of them wrote a single word — they insisted on live dialogue, particular styles of question and answers. When they died, their students wrote down their words and invented these philosophical dialogues. I thought, “Interesting that there’s a new form of dialogue that’s emerging through these chat bots, I wonder what happens if I can use the dialogic form of the chat to access dialogic philosophers like Socrates?” You upload a defined data set — let’s say the Platonic Dialogues — and then you generate instructions. Through trial and error, I figured out how to shape that combination of data set and instructions so that you can talk to Socrates. In the instructions, there’s a lot of, you know, “Don’t do this,” “Don’t do that.” You basically have to define what it means to be Socrates and speak. Because Socrates hadn’t written down his dialogues, I had to say, “Refer to the dialogues,” but say, “In a conversation later recorded by my student Plato, I observed …” I wanted it to be concrete and give actual quotations from the text and their answers. It’s fun to chat with these figures. I think you actually really learn something from them.

So how worried or optimistic are you about AI’s potential to impact the humanities?

Silicon Valley is so futurist in its predictions, I admit to being sometimes susceptible to that, just because almost everyone else in the arts and humanities is so knee-jerk against all of that. But I am aware that the utopian promises have not always come to pass and that’s certainly going to be the case with AI as well. Technologies are tools and it’s good to learn how to use them well. They have their possibilities, and they have limitations. I think people feel a lot of fear; we have all these science fiction scenarios in our minds that are often paired with a kind of contempt. “This is all it can do?” “This is not real art.” I think we need to move beyond that, I really do.