Have you heard? AI companies are hiring philosophers!

Sort of.

News about jobs for philosophers is the kind of thing you might expect to read about at Daily Nous—for example, here is a post about which philosophers are working for which AI-related firms and organizations—yet articles about the topic have been cropping up lately in a surprising number of mainstream media outlets.

The picture such articles have tended to paint is that the development of artificial intelligence has brought with it a new career path for people trained in philosophy. But how accurate is that picture?

In reality, things are more complicated and less clear, as Aaron Kagan explains in the following guest post. Kagan is the founder of GraspingAI, an AI research and strategy organization. He holds a PhD in philosophy and has worked at Google Search, X / Loon, Meta Reality Labs, and other technology firms. He is also co-author of An Introduction to Embodied Mind: Thinking Outside the Head and serves as Chair of the American Philosophical Association’s Committee on Non-Academic Careers.

UPDATE: Dr. Kagan says that he used AI as an editorial aid during the drafting and revising of this post, but that “the argument, research, job-posting analysis, and conclusions are my own.”

The Philosophy Job Market Deepfake
by Aaron Kagan

AI has produced a strange new fantasy about the humanities: maybe philosophy was the practical major all along.

In the past few weeks, the story has been hard to miss. Wired told readers that to land a job in AI, they might try reading Kant. The Atlantic announced that someone finally wants to hire philosophers. Business Insider argued that AI is opening doors for philosophy majors. The Economist asked why big AI labs are hiring so many philosophers. International Business Times said AI companies were “recruiting philosophers.” And now The New York Times has given the story its biggest platform yet, under the headline “The Revenge of the Philosophy Majors.”

The message is irresistible, especially at a moment when students, parents, universities, and laid-off workers are all trying to make sense of a brutal job market. AI may be destroying some jobs, but perhaps it is creating a new home for people trained to think about values, meaning, minds, and morality.

I understand the appeal. I have a PhD in philosophy. I spent years inside major technology companies, including Google, X, and Meta. I have also spent years helping philosophers and humanists translate their training into non-academic work. I would love the optimistic story to be true.

But it is only half true. And that makes it dangerous.

The better way to say it is this: philosophy is becoming more visible in AI, but not yet legible as a career path. A few philosophers have highly visible roles. Many more people are doing philosophy-adjacent work under other titles. But that is not the same as a robust labor market for philosophers, and it is not the same as a reliable pathway for students or displaced academics trying to understand where their training fits. The pipeline is real at the level of problems. It is weak at the level of occupational labels.

AI companies really do need work that overlaps with philosophical training. Their leaders increasingly talk about values, judgment, alignment, public input, human goals, what systems should refuse, and what it means for AI to act well in the world. These are not minor side issues. They are conceptual problems that engineering alone cannot settle.

But when you look at the job market, the picture changes.

In a June 25 snapshot that philosopher Charles Lassiter and I conducted, we examined 1,815 currently open roles across 11 AI labs. None required a philosophy credential. A naive keyword count made the market look much larger: 26.6 percent of postings mentioned AI ethics, safety, alignment, governance, or policy. But after removing generic mission language and other boilerplate, roughly 5 percent of roles substantively involved that work. That is a real slice of the market, but a small one, and it is not the same thing as a pipeline for philosophers as philosophers.

The roles are rarely called “philosopher.” They are called research engineer, policy analyst, data scientist, trust and safety specialist, governance lead, UX researcher, product manager, AI safety researcher, legal expert, or societal impacts scientist. The questions may be philosophical, but the job labels usually are not.

The mistake is the old “major equals job” fantasy in a new AI costume. Philosophy does not usually become a job called “philosopher,” just as history does not usually become a job called “historian.” It becomes a form of training that has to be translated into institutional work.

That is the job market deepfake: a public story that turns scattered, specialized roles into the appearance of a clear new pathway.

This matters because the tech job market is already confusing enough. Entry-level workers are being told they need senior-level judgment. Mid-career workers are being told to “learn AI.” Students are being told that degrees, prestige credentials, and traditional entry-level pathways no longer mean what they used to. Then comes a beautiful story: the people who studied Kant, Aristotle, ethics, consciousness, and political theory are finally wanted in Silicon Valley.

Some are. A small number of philosophers really are working inside frontier AI labs. Amanda Askell at Anthropic is the most visible example, and her work shaping Claude’s behavior is genuinely important. Henry Shevlin recently joined Google DeepMind with the actual title “Philosopher,” working on machine consciousness, human-AI relationships, and AGI readiness. These examples matter.

But a handful of high-profile philosophers is not the same thing as a robust, legible labor market.

The difference matters for the student whose family forwards them one of these articles and says, “See? There are jobs for you after all.” It matters for the humanities department trying to defend its relevance. It matters for the laid-off researcher, policy worker, or academic who is told that AI companies are hungry for exactly the kind of conceptual training they have spent years developing.

The truth is more awkward. AI organizations need conceptual labor, but the public labor market does not yet show a clear pathway for philosophers as philosophers.

By conceptual labor, I mean the work of clarifying what a system is for, what assumptions are hidden inside a metric, what values are being traded off, what counts as harm, what kind of evidence is relevant, whose perspective is missing, and what a product should refuse to do. This is not merely “ethics” in the narrow compliance sense. It is the work of making confused problems thinkable. It is one thing a philosophical education can train you to do.

Any serious AI organization needs this work. But it rarely appears as a standalone occupational category. Instead, it is often routed through categories organizations already understand: engineering, policy, law, data science, product, and trust and safety.

That translation is not always bad. A trust and safety analyst deciding when a system should refuse a request may be doing applied ethics under another name. A product researcher studying how people understand an AI assistant may be doing philosophy of mind in practical form. A policy worker defining “misuse” may be doing conceptual analysis with operational consequences.

The problem is not that philosophy is absent from AI. The problem is that it often becomes invisible inside the organization.

I have seen this firsthand. Years ago at Google, I helped surface and organize an informal “Philosophers at Google” group. It began with a small call: who else here has a philosophy PhD or serious philosophical training? A few of us met. Then more people surfaced: product people, policy people, UX researchers, engineers, strategists, and people working in AI ethics and safety-adjacent roles.

Eventually, the group grew to around thirty people. The funny thing was that none of us had been hired as “the philosopher.” We had translated ourselves into different roles the company understood. But once we found each other, it became obvious that the same kinds of training kept reappearing: clarifying concepts, surfacing assumptions, building taxonomies, asking what problem was actually being solved, and helping teams reason about human consequences.

The talent was already there. The institution just did not have a clear way to see it or name it.

That is why the current “AI wants philosophers” story needs a correction. The issue is not whether philosophers can contribute to AI. They already do. The issue is whether the public story has mistaken philosophical demand for a legible occupational pipeline.

A better headline would be: AI companies say they need philosophers. Their job boards tell a messier story.

That may sound less romantic than “read Kant, get hired.” But it is more useful. It tells students and workers the truth: philosophical training can matter in AI, but it probably will not be enough on its own. You will likely need to translate it into product judgment, research practice, policy design, safety work, governance, law, or technical collaboration. The work is real. The pathway is indirect, uneven, and often hard to see.

And it tells AI companies something they need to hear: if values, judgment, public input, and human goals are truly central to the future of AI, then they cannot remain vague public language. They need names, roles, decision rights, and career paths. They need people who are explicitly asked, authorized, and rewarded to ask the questions that engineering alone cannot answer.

The danger is not that AI companies are lying when they say they need philosophy. The danger is that the public hears “jobs,” while the labor market says something murkier: maybe, but only if the work can pass as something else.

That is not a pipeline. It is a mirage with a good headline.