尤物视频

Global Academic Publisher

Sophie Goldsworthy

“AI has not rescued the humanities. What it has done is reveal, with unusual clarity, what happens when societies allow the institutions that cultivate judgement to erode.”

For a decade, the standard advice to a philosophy undergraduate was to have a backup plan. Now, several of the world’s most consequential technology companies have hired philosophers to do work that sits close to the centre of what they build, not at the edges of it. At Anthropic, philosopher Amanda Askell 鈥� lead author of Claude鈥檚 鈥淐onstitution鈥� 鈥� spends her working days training AI models 鈥渢o be more honest and to have good character traits.鈥� At Google DeepMind, philosopher Iason Gabriel is leading research into the 鈥渃omplex moral landscape of AI鈥�.

OpenAI haven鈥檛 made hires, but claim to have consulted hundreds of moral philosophers and ethicists in shaping ChatGPT鈥檚 guardrails, and Sam Altman acknowledges the influence of the 2014 book Superintelligence 鈥� widely regarded as the origin text on which current work on AI safety and guardrails builds 鈥� pronouncing it 鈥渢he best thing I鈥檝e seen on this topic鈥�. As concluded, 鈥渙ne of humanity鈥檚 oldest disciplines and one of its newest inventions feel distinctly made for each other.鈥�

This reads, at first glance, like a rescue story. What has actually happened is narrower and more interesting. AI has made a specific set of humanities capabilities newly legible to employers who had no prior use for them. Reasoning about truth, belief, obligation, and interpretation used to be a hard sell outside a seminar room. It no longer is, in a handful of places, for a handful of people. As Aaron Kagan described it in the philosophy blog , this is 鈥渢he job market deepfake: a public story that turns scattered, specialized roles into the appearance of a clear new pathway.鈥�

The industry also has a documented habit of treating this kind of work as disposable when conditions change: ethics and responsible-innovation teams have been cut at Twitter, Meta, Microsoft, and Google within the past three years. Palantir CEO Alex Karp has gone further, predicting at Davos in January that AI 鈥渨ill destroy humanities jobs鈥�. A handful of current hires proves demand exists, not that it will last.

Challenges remain

The humanities face a structural threat from enrolment decline, funding stress, course closures, and weakening pre-university pipelines. The reported in 2025 that degrees in most humanities disciplines fell by more than 25% over the last fifteen years. The says SHAPE subjects lost some 4,000 academic posts in a single year, with early-career staff and teaching roles particularly exposed.

The British Academy also warns of 鈥渃old spots鈥� where languages, classics, and other subjects are becoming unavailable, with this geography of opportunity narrowing fastest in rural, coastal, and historically under-served regions. More than half of young UK students now study within 90km of home, and almost a third within 30km. Students from disadvantaged areas are even more likely to study locally. When regional provision disappears, the choice is lost first by those least able to move.

This is where the AI-hiring story becomes more than a curiosity. The question is not whether AI companies occasionally hire philosophers. It is whether the students who might become the next generation of AI ethicists, linguists, historians, or philosophers will have the opportunity to study those subjects at all.

Access to study humanities disciplines is shrinking

I have a vested interest here: I have spent my career publishing humanities research, at a press that has invested in these subjects over generations, often in fields whose value was obvious to scholars long before it was obvious to the wider world. That record is old, stable, and deliberately slow in the best sense: not inert, but built to outlast fashion, institutional churn, and the argument of the week. It is infrastructure rather than decorative inheritance: once the expertise, teaching capacity, and research communities behind it disappear, they are difficult to rebuild. The British Academy’s report makes the same point in policy terms: hundreds of individual course closures can amount, in aggregate, to a system-level failure.

And when AI tools level the technical playing field, the skills becoming most essential 鈥� well beyond the academy 鈥� are human judgement, interpretation, ethical reasoning, historical perspective, and language. Knowing what to trust, what an argument actually establishes, what a piece of AI-generated text is tacitly assuming, and where fluency is concealing ignorance. Large language models can generate answers in seconds. They cannot, by themselves, tell us whether those answers are well founded, ethically defensible, historically informed, or rhetorically misleading. The phrase 鈥淎I slop鈥� has caught on because it names a real fear that synthetic abundance will make knowledge feel cheaper, faster, and less accountable. Scholarship is not valuable despite being careful, slow, and evidenced, but precisely because of those qualities.

But access to that scholarship as published content cannot compensate for the loss of access to the disciplines themselves. The British Academy report is clear that shrinking regional provision reduces student choice, deepens barriers to opportunity, and threatens the research pipeline that sustains future expertise. Languages provision has nearly halved since 2011/12. English Language and Literature enrolments have fallen by almost half; History and Drama are each down 29 per cent. These numbers are less a lament for the past than a warning about the future: students do not choose from an abstract national menu of options, but from what they were taught at school, what they can afford, where they can travel, and whether the subject still exists within reach. Falling demand may therefore be less a verdict on value than the visible end-point of restricted opportunity.

That is why the debate about philosophy graduates in AI companies is too small, even when it is interesting. A chatbot can summarize Aristotle. It cannot teach a student why a translation matters, why a premise is doing more work than it admits, or why a confident answer may still be wrong. Nor can it rebuild a department once the staff have gone, the courses have closed, and the local pathway into a subject has disappeared.

Humanities skills will remain vital

A small number of new job titles do not settle the future of the humanities. What they do is put a number on the clock. The pipeline that produced those philosophers stepping into roles at leading technology companies is narrowing, while the need for their skills is only starting to reveal itself. Whether there are more philosophers available to step into these roles in 10 years depends on decisions being made now, and not by AI companies at all. It depends on schools that still teach students how to read slowly. It depends on universities that still make room for languages, history, philosophy, literature, and the study of culture. It depends on publishers, libraries, and funders continuing to sustain the scholarly record from which all serious interpretation begins.

AI has not rescued the humanities. What it has done is reveal, with unusual clarity, what happens when societies allow the institutions that cultivate judgement to erode. The skills that help us distinguish evidence from assertion, interpretation from prediction, and knowledge from plausible-sounding noise are becoming more valuable at exactly the moment access to the disciplines that teach them is becoming more fragile. Philosophers Amanda Askell, Iason Gabriel, and several of their colleagues had that chance. The real question is who still gets it after them.