From “Editor” daguerreotype (ca. 1855) | Unidentified Artist / Smithsonian American Art Museum, Gift of Charles Isaacs / CC0
Rather than focus on techno-utopian fantasies or doomsday predictions in which technology replaces humans, scholar Dennis Yi Tenen inspects writing itself as a human technology. In his new book, Literary Theory for Robots (W. W. Norton, 2024), the English professor and former Microsoft engineer asks: How will AI change the technology of writing. Recently, Yi Tenen sat down with Tsering Dolka Gurung for a conversation about the writing technologies from which AI developed, such as Buddhist divination practices. Literature, Yi Tenen says, will always be a human game played and watched by humans. We are not interested in robots—we are interested in each other.
Tsering Dolka Gurung: Can you talk about how your educational and work experience informed this book?
Dennis Yi Tenen: My undergraduate and graduate education is in literary studies. And then I worked as an engineer, and because of that, I have an engineering approach to things. Say I’m disassembling a toaster. I ask, “How is it made?” When I look at a book, I also ask, “How is it made?” My background in engineering makes me ask how something was constructed. Engineering is a question of poetics.
Gurung: What sparked your interest in the history of artificial intelligence? Anger? Frustration?
Yi Tenen: Good question. I am frustrated with people assigning AI agency. If you look under the hood, you see that it’s a constructed entity, made by humans—made by teams of humans. Rather than an existential threat, AI is the result of a long history of collective human labor.
Gurung: One of your main arguments is that artificial intelligence is irreducibly human. In what ways do AI systems reflect human thinking rather than independent intelligence?
Yi Tenen: You can go to the library to converse with dead authors. With the advent of a search engine, you don’t have to physically go to the library. A search box gets you to the specific place in the archive more quickly. AI is an acceleration of that. It’s an interface to the library archive on which it was trained. When the technology summarizes for you, it’s giving you the scientific consensus on a topic. That consensus wasn’t the product of AI, but of long-standing human collaboration.
Gurung: Will AI change the way we value literature?
Yi Tenen: Humans play human games because we love drama. Do you want to watch football played by robots at supersonic speeds? No, you want to watch football played by humans who are trying, failing, shouting at each other, doing victory laps. We’re not interested in watching machines perform. We’re interested in each other.
Gurung: If a robot were to write something that moved people to tears, does that mean the robot understands sadness?
Yi Tenen: Humans are interested in watching human games. Literature is a human game. And the games of literature also follow pretty generic rules, or templates. Perry Mason, a very long-running and popular show, began as a series of novels, sixty or seventy novels, written by this author through algorithms. He was using templates. Human creativity involves reuse. It involves algorithms. It involves templates. Templates don’t “understand.” Neither does AI. Using templates to generate human emotion is a human practice, but the templates themselves are inert.
Gurung: People think AI threatens the role of human writers. Others say it enhances creativity. Where do you stand?
Yi Tenen: Technology changes what creativity is. It doesn’t substitute for creativity. Twenty years ago, you would hire a wedding photographer for weddings. Now everybody has a camera on their phone, it’s super smart. It takes incredible pictures. When iPhone camera technology was first introduced, people thought, “Oh, my God, professional photographers are done for! Everybody has a phone camera!” Well, that’s not true. Photographers do very well today, just like before. Yes, what it means to be a photographer today is very different. You can no longer just show up with a camera and press the button. Now, photographers may be flying a drone, employing sophisticated editing tools. What it means to write will change. I think it’s a positive trend, because the profession becomes more interesting, more sophisticated. We often romanticize that old-school labor. And that’s fine! I love notepads. But it’s exciting to imagine writing as a much more sophisticated, creative enterprise.
Gurung: How do you think AI-generated writing will evolve?
Yi Tenen: For the author of the future, writing will involve an orchestration of resources. A contemporary film director, for example, is not doing everything in the film. She hires a team. She orchestrates. “You do this. You stand over there, this camera moves like this.” There are post-edits. Writing was sequentially putting pen to paper. But writing tomorrow will look more like directing a film. You’re orchestrating resources, generating something, putting it through one filter, applying another filter, tuning stuff. An image is a signal. Writing is also a signal that you’re putting through a number of transmutations to produce your art.
Gurung: Bill Gates said that in the next ten years, AI will completely replace humans. We will work only two to three daily. What do you think about that?
Yi Tenen: In the nineteenth century, people worked in unsafe conditions for twelve hours per day, and now we have an eight-hour workday. We have more leisure time. But it’s not like we get to completely relax. Whether it’s world hunger or the environment, or making our society a better place, there’s stuff to do. You can’t just go home and take a nap. If you were given infinite vacation, you would get bored. Sure, you’re on this beautiful tropical island taking a vacation. But at some point, you’ll say, “Ah, let’s make it better!”
A good analogy here is a calculator. If you told me to calculate a square root by hand, I couldn’t. I don’t remember how to do that. I can do it on my phone in one second. But did I even have any natural capacity to begin with? In the book, I suggest without tools I don’t know anything. Most of my creativity and my knowledge is social. I was told there is something called an electron. But fundamentally, I don’t know. I feel smart because I’m surrounded by books. I’m surrounded by people. Knowledge has always been like that—it doesn’t replace us. Plato criticized books for giving readers the appearance of knowledge by looking stuff up. But does that mean we’re gonna ban books? No!
Gurung: I’m a Buddhist, and in our tradition we have a form of divination called “Mo,” which we use to make predictions about travel, health, or work. In your book, you connect such divination practices—like that used by Ibn Khaldun—to AI. Could you talk a little more about that?
Yi Tenen: All cultures have oracles. And in a way, they’re proto-algorithmic. Oracular devices are a way to manipulate objects and manipulate charts to produce meaning. Divination, generating text automatically, is as old as writing. Zairajah charts, for instance, were used to create pulp fiction in the 1920s in the United States. They were quite similar to these medieval divination circles, but people were using them to write pulp fiction and science fiction and detective novels. And then with the first generation of computers in the in the late 1950s, early 60s, came the desire to automate the generation of text. The first thing researchers did was put those paper machines into code. When we talk about magic and machines here, it’s not just an analogy. There is a lineage between the divination devices of Buddhism and Islam and Kabbalah, between the medieval rhetorical automation and the writing practices of mass-culture (pulp fiction, Broadway, and television). This technology seems magical because AI is an acceleration of those divination practices. It’s deeply familiar. And that’s why we shouldn’t be surprised that it seems magical, because it was always magical.
