This seems as good a moment as any to recall the so-called Sokal Affair. The Spring/Summer 1996 issue of the journal Social Text included an article, titled Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity. As per Google Scholar, the article currently has over 2000 citations. The author of the article, Alan Sokal, emeritus Physics professor at New York University, subsequently claimed, in a widely read note, that the article was what we would call “slop” in the modern-day.
In an attempt to show a “decline in the standards of intellectual rigor in certain precincts of the American academic humanities”, Sokal wanted to test whether an article “liberally salted with nonsense” — his primary concern was with the use of “scientific and mathematical concepts in ways that few scientists or mathematicians could possibly take seriously” — would pass muster with the editors of a “leading North American journal of cultural studies” if “(a) it sounded good and (b) it flattered the editors’ ideological preconceptions.” It did.
“Poor Sokal.” Thus wrote Jacques Derrida, one of the major figures associated with cultural studies — Derrida pioneered deconstruction, a particular way of approaching texts, which supplied the methodological foundations for the field — who was subject to a lot of negative press, especially in US newspapers, after the publication of the article. Sokal, Derrida argued, had chosen cheap tricks over a serious engagement with his ideas, noting that he was “always sparing and prudent in the use of scientific references.” Derrida’s admonishments notwithstanding, it was cultural studies that emerged as a lesser force after the Sokal affair, its credibility considerably diminished.
Thanks to generative AI, cultural studies may finally have an answer to the challenge that Sokal had posed three decades ago. At least, that is what Jordana Cepelewicz, Maths editor for the Pulitzer Prize-winning web magazine Quanta seems to suggest, in a recently published article on the platform, titled Is AI the End of Math As We Know It?.
The crux of the article is that OpenAI’s recent publication of proofs for hundreds of previously unsolved mathematical problems had disrupted the very foundations of mathematics as a discipline because of its obsession with problem-solving and theorem-proving. The article contends that OpenAI’s discoveries meant that the discipline had to reorient itself toward a more interpretative approach, making sense of AI proofs of theorems rather than trying to prove them from scratch.
Also read: India’s IT giants rode labour arbitrage for 30 years. AI is hurting billable hours business model
Navigating ‘language games’
That the field hasn’t taken well to this is perhaps best manifested in the dramatic breakdown of Steven Strogatz, a Maths professor at Cornell University, over a New York Times podcast. Strogatz went so far as to say that for many in his fraternity, solving theorems gave a “sense of purpose in life” and it grieved him to see that snatched away by gen AI. Cepelewicz, however, goes much further, when she observes that she was “reminded of [the literary theorist] Roland Barthes’ famous 1967 essay ‘The Death of the Author,’ in which he argues that it is the reader, and not the writer, who has ultimate control and agency in actively creating a work’s meaning”.
To see why this is simply an astonishing statement, we need to return to Sokal’s explanation for why he decided to try his “experiment.” Social Text’s acceptance of his article, he argued, “exemplifies the intellectual arrogance of Theory – meaning postmodernist literary theory – carried to its logical extreme…If all is discourse and ‘text,’ then knowledge of the real world is superfluous; even physics becomes just another branch of Cultural Studies.” Worse still for Sokal, since “all is rhetoric and ‘language games,’ then internal logical consistency is superfluous too: a patina of theoretical sophistication serves equally well.”
While it is too early to say that mathematics, or the mathematically inclined sciences, more generally, will eventually become “a branch of the Cultural Studies,” there should be little doubt that following OpenAI’s extraordinary revelations, researchers in these fields will be spending a lot of time interpreting “text.” Now, of course, there ought to be conventions for doing “good” or “bad” interpretation — Sokal is certainly right that his article should not have been accepted by Social Text — but the “language games” he saw as the exclusive provenance of the humanities is now upon the natural sciences as well.
Moreover, if natural scientists cared to look, there is lots of advice accumulated by the socially oriented sciences on how to navigate these “games.” The important question, as Cepelewicz correctly posits, is whether young mathematicians, raised in higher education systems that have elevated theorem-proving to a higher calling, will be willing to undertake the less celebrated task of interpretation, or vote with their feet.
Subhasish Ray is Professor and Associate Dean of Research at the Jindal School of Government & Public Policy, O.P. Jindal Global University, Haryana. He tweets @subhasish_ray75. Views are personal.
(Edited by Aamaan Alam Khan)