The Humanities in the Age of Generative AI

The humanities are once again being declared obsolete. With the advent of generative artificial intelligence, particularly large language models, commentators have begun to ask whether humanistic skills are still needed at all. If machines can write fluently and critique persuasively, what remains for disciplines premised on interpretation, argument, and, most notably, the (often intentionally) slow cultivation of insight? These questions are not new, and the humanities have faced successive waves of scepticism over the past century: the utilitarianism of post-war science policy, the expansion of vocational higher education in the 1980s and 1990s, the rise of data-driven research paradigms associated with the digital turn, and most recently, the claim by Palantir CEO Alex Karp that AI will ‘destroy’ humanities jobs (Munis 2026). Each time, the question has been posed in similar terms: what use are the humanities in a world preoccupied with speed and measurable (typically economic) outcomes? The emergence of generative AI represents something more than the latest ideological pressure. Since the post-war era, pressures on the humanities have largely been a function of institutional incentives that advantage STEM disciplines (Small 2013), but the threats posed by generative AI feel different, and they are different, because they fundamentally shift the structures of knowledge production. Some subfields, ethics and philosophy of mind among them, have clearly gained some ground as AI has become a matter of public concern. The humanities are not uniformly in retreat, but the cumulative effect of these structural pressures, together with the capabilities of generative AI, produces a threat to humanistic inquiry that exceeds what any single one of them would produce on its own. Earlier pressures operated largely on the humanities’ institutional footprint, on resourcing and prioritisation, while leaving the underlying intellectual activities intact. Large language models operate on writing, the medium through which that activity is conducted, which is why it warrants a different kind of attention even from scholars who have lived through several previous declarations of crisis. Large language models significantly reshape the processes by which textual meaning is generated and legitimised, and their implications are epistemic as much as pedagogical or institutional. The challenge, then, is not how the humanities can survive generative AI and large language models, but how they can respond on their own terms, by interrogating the assumptions and politics of these technologies. Generative AI is a threat to humanistic authority, but the response to that threat lies in re-articulating that very authority in dialogue with the technical systems that increasingly mediate our intellectual life. Humanities graduates and humanities-trained researchers already contribute to model development, policy formation, and applied ethics work inside technology companies, and that contribution is neither denied nor displaced by what follows; the question addressed here is the narrower one of how universities, as the institutions that still train most humanists and still produce most humanistic research, should think about teaching and learning, research, and governance under conditions in which generative AI has become infrastructural.