An English Professor Confronts AI

Photograph Source: Michael Cordedda – CC BY 2.0

What is a hard-working English professor to do when confronted with the fact that ChatGBT can write perfectly acceptable undergraduate papers interpreting literary texts?  Are we out of a job? Can our students cheat their way through college without ever learning how to read a poem or a novel?  And who really cares anyway?  Haven’t students always done this, copying large parts of their papers out of an encyclopedia or Wikipedia?  Shouldn’t students be concentrating on practical matters, preparing for jobs that don’t require a college degree, or majoring in one of the STEM disciplines, Science, Technology, Engineering, Math?

As way of putting my toe into these deep waters, I tried an experiment in my course on William Blake at the University of Chicago this spring.  I asked the students to pick out a “Proverb of Hell” from Blake’s The Marriage of Heaven and Hell, and to ask ChatGPT to provide a one-page interpretation of it.  They were instructed then to write a critical commentary on the results, looking for mistakes, blind spots, and opportunities to make a better interpretation.

The first response was both reassuring and disquieting.  ChatGBT turns out to be a rather good interpreter of complex and enigmatic proverbs.  In response to “Sooner murder an infant in its cradle than nurse unacted desires,” the Bot instantly realized that the proverb was not to be taken literally, instead reading the infant as a metaphor for desire, and linking it to Blake’s other statements about the need for expression to overcome the tendency of human beings to repress their feelings.  The student who picked this proverb found that the only way to critique the Bot was to revive the literal reference to a murdered infant as a shock to the sensibility of normal readers.  After all, this is “the Voice of the Devil,” not the rational Angel, so its status as an “excessive” statement needs to be registered.  This is in line with another proverb that tells us “The road of excess leads to the palace of wisdom.”  In short, the statement is not merely metaphor, but also hyperbole.

Another student asked the Bot to interpret the proverb, “Drive your cart and your plow over the bones of the dead.”  AI produced a perfectly reasonable paraphrase, advising us not to dwell on the past, but focus on the present and future.  The student pointed out that this air of reasonableness ignores the long-standing need of human beings to honor and remember the dead.  She criticized the Bot for failing to register the emotional horror of the proverb, turning it into a flaccid cliché of common sense—a perfect example of mechanical thinking. [1]

I asked the Bot to interpret my own favorite proverb, “If the fool would persist in his folly, he would become wise.”  The answer was quite reasonable, noting that human beings learn by trial and error, and even foolish actions can help us learn by experience.   I provided my own critique, pointing out that sometimes people with bold, original ideas are regarded as fools, but if they persist long enough people may come to realize that the idea was actually wise and powerful.  My example was the long-standing dismissal of Artificial Intelligence itself as a fool’s errand by many experts in computer science.  When I communicated this interpretation to the Bot, it responded that this was an interesting reading, and it would incorporate it into future requests to interpret this proverb.

In the last few months, AI has been second only to the debt crisis as a persistent obsession of the New York Times.  Since the publication of journalist Kevin Roose’s dialogue with ChatGPT back in February, in which the Bot “fell in love” with the journalist, urging him to leave his wife and “do love” together, scarcely a day has passed without a story on the threats and promises of artificial intelligence.  The spectre of killer robots and Dr. Strangelove’s automated weapon systems is haunting the world with the prospect of the human species destroying itself with its own technologies.  Dr Geoffrey HInton, the “godfather of AI,”  has been re-enacting the role of Dr. Frankenstein, warning us that he may have created a monster.  “Easy access to AI text- and image-generation tools could lead to more fake or fraudulent content.”  He has called for a six month moratorium on AI research. The average person, he warns, will “not be able to know what is true anymore.”[2]

But even above average persons like the readers of CounterPunch may have some difficulty knowing what is true. This has been the condition of the human species since we first tasted the fruit of the Tree of Knowledge.  Still, it is welcome news to humanists that scientists and engineers (and even a few capitalists) are thinking about the ethical implications of their inventions.  But a six month pause for reflection is not nearly long enough.  Better to build history, culture, and moral philosophy into the STEM disciplines as a permanent feature.  We could then develop programs in what we will call the “SCHTEM” disciplines (Science, Culture, History, Technology, Engineering and Math) and we humanists can help our machine obsessed colleagues ponder the relation of the human species to its inventions in a systematic and sustained way.  Meanwhile, we English professors could stop obsessing about the threat of plagiarized freshman themes and encourage our students to treat AI as a friend and collaborator.   This will inspire their competitive instincts and help them become smarter than the machines that are generating and interpreting all these words and images.  Remember that those images and words could be, not just fake selfies and their accompanying lies, but fraudulent satellite images and commands to launch a first strike.  If the humanities are the discipline that reflects on human nature and the fate of our species, perhaps it might help us to stop using our gadgets to render us extinct.

One thing seems certain:  the whole question of exactly what intelligence is has been opened for fresh investigation.  How does human intelligence measure up against machine, animal, and even vegetative intelligence? (See Eduardo Kohn, How Forests Think). What are the criteria that allow us to judge who or what is smarter?  One answer is clearly evolutionary success, defined as durability, longevity, adaptation, cleverness, inventiveness, skills, and cooperation with other species.  High on the list of the dubious achievements of human intelligence is our ability to destroy other species along with our own habitat.  Our much-vaunted “creativity” is all too often a form of what Joseph Schumpeter called “creative destruction,” the key feature of deregulated capitalism and its partner, environmental devastation.  Why not bring our English majors into conversation, not only with the STEM disciplines, but with economics and ecology, and the new forms of machine intelligence that command the entire archive of human and natural history?


[1] See Stephen Eisenman’s article, “AI Chatbots Are Even Scarier Than You Think,” March 3, 2023:

[2] Fortune Magazine (May 1, 2023):

W. J. T. Mitchell is Professor of English and Art History at the University of Chicago and Senior Editor of Critical Inquiry. He is the author of numerous books on media, culture, and politics.