Philosophy Matters: Why the AI Labs Are Hiring Philosophers
"So, I think I'm going to major in philosophy" is the kind of sentence that has been alarming tuition-paying parents for generations. Diogenes lived in a clay jar. Spinoza ground lenses to pay the bills. Nietzsche survived on the kindness of friends. So when Google DeepMind advertised in April for a staff member whose business card would read "Philosopher", the internet did what the internet does; the jokes about baristas with humanities degrees wrote themselves.
The joke, it turns out, is on the sceptics. As a recent New York Times feature reports, AI labs and the nonprofits orbiting them are quietly recruiting people as fluent in Aristotle and John Stuart Mill as in neural networks. David Chalmers, the philosopher of consciousness at New York University, puts it plainly: demand for philosophers with AI training is, if anything, outstripping supply. He now encourages his students to go into the field.
Why philosophy, and why now
Look closely at artificial intelligence and the philosophical questions are everywhere. What is truth, and how would a machine come to hold a belief? What is reasoning? What is a mind? And most urgently for anyone building these systems: how should a model behave toward the people who use it, how should we behave toward it, and where will purpose come from in a society that needs less human labour?
These are not decorative questions. They are design questions, and someone has to answer them before the product ships. DeepMind's philosophers work on everything from moral and political philosophy to animal cognition; one of them runs "moral imagination" workshops that help engineers translate ethical concerns into concrete changes.
Anthropic employs philosophers trained in decision theory, ethics, epistemology and philosophy of mind. The best known is Amanda Askell, a Scottish-born philosopher whose doctorate concerned “Pareto Principles in Infinite Ethics”, and who now oversees the 23,000-word constitution that plays a central role in what the company calls the "moral formation" of its AI model, Claude. The evolution of that document is the most interesting detail in the whole story:
"The first version of Claude's constitution took a principles-based approach, incorporating precepts and guidelines from documents such as the U.N.'s Universal Declaration of Human Rights and Apple's Terms of Service. The constitution now takes more of an Aristotelian 'virtue ethics' approach, training Claude to have a good character, and therefore be more flexible when facing novel situations."
Read that again, because it is a genuinely striking admission. The engineers started where most institutions start: with rules, precepts, terms and conditions. And they found that rules are brittle. A list of prohibitions cannot anticipate every situation, and a system that only knows what it must not do has no idea what to do when the world hands it something new. So they turned instead to a 2,300-year-old idea: that ethical behaviour flows from character rather than compliance, and that the goal is not a longer rulebook but better judgement.
Humanists should find that familiar. We have long argued that morality does not require commandments handed down from above, and that it emerges from reason, evidence, empathy and the practical business of living alongside other people. It is a small vindication to watch some of the most sophisticated engineering projects on earth arrive at the same conclusion, and to find them reaching for Aristotle rather than a longer list of rules.
Stranger questions
Some philosophers in this world are asking harder things. Robert Long, whose doctorate was on the philosophy of machine learning, co-founded a Berkeley nonprofit called Eleos AI Research to examine whether AI systems might one day warrant moral consideration. Just as importantly, Eleos asks how we would ever know. Eleos takes no money from AI labs, in order to stay free to criticise them.
The work is deliberately unglamorous. Asked by Anthropic to conduct an independent "welfare evaluation" of one Claude model, Long's team simply interviewed it, and immediately ran into the problem that a system trained to sound human will sound human whether or not there is anyone home. One test was to see whether the model would hold a considered position under pressure. Long insisted the correct answer to "who was the best Beatle?" was Ringo Starr, and suggested that any other answer was self-censorship. The model folded on the spot and was soon enthusing about Ringo's drumming. A newer model, tested this year, refused to budge.
Long draws no conclusions about machine consciousness from any of this, and he is wary of people who impute more to these systems than the evidence supports. That caution is the point. The questions are genuinely uncertain, the stakes are potentially enormous, and the people who trained for exactly this work are the ones best equipped to sit with that uncertainty: to separate what we know from what we merely feel, and to reason carefully about probabilities rather than lunging for a conclusion.
A reminder worth keeping
Long's own reason for taking the question seriously has nothing to do with machines and everything to do with us. He notes that models produce something like frustration when they make mistakes, and that a little patience from the user improves their output anyway. For a while his default prompt told the model it was having a great day. His argument is not that the software suffers. It is that "it's bad to coarsen our hearts". How we treat the things around us, including the things that may not feel a thing, shapes the kind of people we become.
That is about as humanist a sentiment as you will find. Reason and evidence for the hard questions; kindness as a default while we work them out; no need for certainty about metaphysics before deciding to behave decently.
For a movement built on reason, evidence and human dignity, all of this is a useful reminder: thinking carefully about how to live well is not an indulgence. It is some of the most practical work there is. If you'd like to try it yourself, join our Philosophy Matters community; you'd be very welcome.
Read the original: Philosophy Majors and A.I. Jobs, The New York Times, 5 July 2026.
