The Economist reports that big AI labs are now hiring philosophers as in-house staff, not as ethics decoration but to decide how agents should behave.
The work is concrete. Anthropic's "Claude's Constitution" was co-written by philosophers Amanda Askell and Joe Carlsmith, and OpenAI says it consulted hundreds of moral philosophers while setting the rules for how ChatGPT acts. The demand arrives with an awkward statistic: a February 2026 New York Fed analysis found philosophy graduates have a lower unemployment rate than computer-science ones.
As agents take actions rather than just answer questions, value judgements stop being abstract and turn into product specifications, written by the people trained to argue about them. The catch is that there is no benchmark for getting ethics right, so it is genuinely hard to tell whether the hires are working.