Why AI will and won’t kill the Humanities?? Stephen Marche’s article “The College Essay Is Dead” for The Atlantic leads with the liner title: “Nobody is prepared for how AI will transform academia.”
He is partly correct in this statement, in truth it is more accurate to say that academics are not prepared for how AI will transform academia.
Why AI will and won’t kill the Humanities
Marche’s first point covers the problem of AI generated essays. It is a seemingly relevant question. How can we tell if a student wrote an essay themselves or an AI? But the issue here is that the problem is not new at all.
Really, how can you tell the difference between a paper written by AI, or a paper plagiarized from another paper, or a paper written by someone paid to write a paper by someone else? There will always be methods of cheating.
The arrival of a new method in itself does not signify a threat to education in general. The internet, for example, opened up many new ways of cheating. But it did not lead to the general collapse of grading systems.
Another, much larger issue, is that Marche conflates the humanistic and scientific traditions in strange ways. Marche cites C.P. Snow’s argument that scientific people are ignoring Shakespeare, before opining that he wishes Zuckerberg had read about the regulation of the pamphlet press in 17th-century Europe.
But these are two different things. (As someone who wrote a book called “How Shakespeare Changed Everything,” Marche can be forgiven for being biased in favor of the importance of something like Shakespeare.)
The problem is that the scientific tradition is valued now because it produces knowledge. (If we assume that knowledge is a falsifiable conjecture that hasn’t yet been proven false.) And the humanistic tradition, or the Humanistic Tradition™ as we know it today, is valued because of its attachments to power and prestige in the past.
(As well as its ability to attract tourists in certain European cities.) As Marche suggests, the regulation of communications technologies in the 17th-century is directly relevant to the regulation of communications technologies in the 21st-century.
Marche is right on this point, the adage “Those that fail to learn from history are doomed to repeat it,” being popular for a reason. But returning to the original example, how is Shakespeare relevant to scientists?
Marche’s citation of Snow’s seemingly prosaic point that there is a need for Shakespeare is an assumption without evidence. In reality, Shakespeare is only important because for about 300 years writers in the English language have said it is important (sometimes with a near-religious fervor), not because it is inherently superior to any other artistic work.
Its inherent importance derives from its relevance to the study of the history of 17th-century English literature, as well as the English cultural nationalist project that enshrined it in its current prominence in the Anglosphere.
And while the study of the all above is part of the scientific tradition, the appreciation of Shakespeare is not. (Being both a novelist and physical chemist, C.P. Snow assuredly understood this difference.)
Ironically the real complaint should be that Zuckerberg is ignoring the scientific tradition in favor of the humanities. Marche, and many others who use buzzwords such as “STEM”, conflate the basic technical task of programming with the much different work of a “scientific discipline.
” Programming is really a tool of creation, an art of expression through a language of programming, much closer to learning how to paint than the more difficult task of learning how to reliably contribute to human knowledge.
(And like painting, will be facing a degree of automation by the latest AI.) The point that Marche misses is that Facebook is Zuckerberg’s artistic expression. His human expression unto the world. And much like any well-known popular work of art, Facebook has been critiqued by many.
(Often, rightfully.) But the most persuasive arguments are those that seek to measure Facebook’s harm on society in scientific terms. If Facebook is a scientific experiment through which to benefit human society, it has failed.
With Facebook, Zuckerberg is sitting on top of a king’s ransom of human data. An empire with more citizens than Augustus’ Rome, a subject Zuckerberg apparently enjoys to read a lot about. Facebook could be focused on finding powerful insights on human behavior and how it can be utilized to strengthen present day democratic and cooperative human systems.
But instead Zuckerberg allowed others to harness it for the opposite. (Academic efforts to make use of Facebook’s platform and data do exist, but are a sideshow compared to the core business model.) Zuckerberg’s goal has never been to increase human knowledge, it has been for the expression of his personal brand of humanity, namely that of a more digitally connected world, for profit, on his own terms, and regardless of the cost to everyone else.
(For insight into what Zuckerberg personally thinks of most popular critiques of Facebook, see here.) If you switched out his apparent interest in Caesar Augustus for Shakespeare, what would that change? (Let us also remember that Elon Musk is big Monty Python fan, a fact relevant only for its irrelevance.)
The pursuit of knowledge requires a dose of humility, as to pose a falsifiable hypothesis implicitly entails the acknowledgement that one might be wrong. Humility, unsurprisingly, is not a trait that is positively selected for among those who join the triple comma net-worth club.
Marche claims that “humanistic questions are real questions with real consequences.” But it is more accurate to say that scientific questions are real questions with real consequences, and that humanistic questions have real consequences only because people think they are real questions. (A real question being one with real answers.)
For evidence look simply to the centuries of religious war, which have still not conclusively proved which celestial being rules us all and how they wish to be worshiped, or the many long-winded arguments found anywhere in the world (and especially the internet) that can be boiled down to a mere question of semantics. (Artificial intelligence has already found these quite easy to churn out.)
This seems an overly aggressive point. But what are the humanities anyways? (Emphasis on the lower-case version.) Marche doesn’t ever define the term despite using it about 26 times. I would offer that it concerns expression of human subjectivity — basically the humanities are just humans being humans. Or in another word, art. (Often, many of the social sciences are grouped into the “term” humanities, check Wikipedia for an example. But this is something I am attempting to deconstruct here.)
This may seem like a broad definition. It is true that in the West the humanities have offered different things at different times. In the history of the earliest universities, following the inspiration of the ancient Greeks, it was part of an education of elites in preparation for citizenship and rule.
Modern universities turned the human arts towards the state-supported furthering of national cultures. The history of the last 50 years of the humanities as a discipline, including the so-called “School of Resentment”, exemplify a successful attack on the above but have been accompanied by a failure to offer any concrete definition of an alternative.
Which isn’t a bad thing in itself, art exists and should exist for its own reasons, but the existence of a guild-like system such as tenured professorships requires there to be a specific, valuable process that is worth protecting. (Hint: in the non-humanities, it is the scientific method.)
So what does this have to do with artificial intelligence? A clear result of advances in generative AI will be a variety of media and forms of communication, forms of which many humans will not be able to differentiate between those produced by human and computer creators.
Because the humanities are subjective by definition, this shouldn’t really be a big deal (at least philosophically, ignoring the economic effects of countless jobs that will be lost to automation). People will better be able to find the content they enjoy, a process that has gotten ever easier, beginning before even Shakespeare. Human expression will have a chance to be even more diverse and democratic, albeit with potential new ways to be controlled and directed from above.
But knowledge production remains a difficult thing to automate. Data collection, yes, but the critical thinking that informs the process remains beyond the reach of current artificial intelligence. Marche is right to be worried, but the problem is not how the humanities are adapting to AI.
The problem is the failure of humanities as taught in higher education to show their relative relevance compared to any of the humanities outside of higher education. (Opinion: they are less relevant.) The problem is that the university is for learning and the humanities as a field have struggled to show in a while that they have been involved in knowledge production.
From the perspective of the humanities, artificial intelligence is speeding up the means of production in a field where quality is difficult to judge, or in some cases can’t be judged at all. This is a problem only if you are employed on the conceit that this quality actually exists and that you can judge it.
The scientific tradition represents our most well-known and reliable method of judgment of this quality, and thus is yet safe from AI barring the advent of the singularity. On the other hand, the humanities have been arguing for quite some time about their own methods. It is only natural for AI to add its own voice to the debate. And it will prove to be very eloquent.