The NYT piece argues that AI labs like Anthropic and OpenAI are actively preferring candidates who can 'specify a problem in prose' over LeetCode grinders. It cites Anthropic's model-behavior team being roughly half non-CS majors and Scale AI/Surge hiring philosophy PhDs at unprecedented rates, framing this as evidence that articulating requirements has become scarcer than typing code.
The editorial reframes the story away from a culture-war narrative and toward a mechanism: writing code has become cheap thanks to Claude Code and Cursor, so the scarce skill is now deciding what to build and articulating it precisely. Philosophy seminars happen to train exactly this skill, which explains the hiring shift without requiring any 'revenge' framing.
The article reports CS grad hiring at top 20 tech employers down roughly 30% year-over-year while new role categories — AI trainer, model behavior specialist, eval designer, forward-deployed engineer — are growing sharply. The implication is that the traditional CS-to-SWE pipeline is being replaced by a broader set of roles where domain reasoning and specification skills matter more than algorithmic implementation.
The editorial explicitly rejects reading this as culture-war payback for STEM, arguing that framing obscures the real mechanism. Developers who internalize that specification and evaluation are now the scarce skills — regardless of their degree — will adapt, while those who read it as identity politics will miss the actionable lesson.
The New York Times ran a piece on July 5 with a headline that would have been unthinkable in 2019: "The Revenge of the Philosophy Majors." The reporting tracks a cohort of humanities graduates — philosophy, English, comparative literature, classics — landing AI-adjacent jobs at rates that now match or exceed CS graduates from the same tier of school. Recruiters at Anthropic, OpenAI, and a rotating cast of Series B startups are quoted as actively preferring candidates who can "specify a problem in prose" over candidates who can grind LeetCode.
The numbers in the piece are the interesting part. New CS grad hiring at the top 20 tech employers is down roughly 30% year-over-year, while hiring for roles titled "AI trainer," "model behavior specialist," "eval designer," and the catch-all "forward-deployed engineer" is up sharply. Anthropic's head of recruiting is quoted saying the company's model-behavior team is now roughly half non-CS majors. Scale AI and Surge are hiring philosophy PhDs at rates that would embarrass a mid-tier analytic philosophy department.
The framing that keeps recurring in the piece is that the bottleneck in shipping AI products has moved from "can you write the code" to "can you specify what the code should do, in enough detail that a model can execute it and you can tell whether it did." That is a job description that maps almost exactly onto what a philosophy seminar trains you to do.
The temptation is to read this as a culture-war headline — humanities revenge, STEM comeuppance, the return of the liberal arts. That reading misses the actual mechanism, and the mechanism is the thing developers need to internalize.
Writing code has become the cheap part of the pipeline. A competent generalist with Claude Code or Cursor can now produce a working CRUD app, a working scraper, a working eval harness, in an afternoon. The scarce skill is no longer typing the code. It is deciding what to build, articulating the constraints precisely, spotting the failure modes before they ship, and writing the specification tightly enough that the model doesn't hallucinate its way into a plausible-looking wrong answer. Every one of those skills is a writing skill first and a coding skill second.
The philosophy majors are winning because their training happens to match this new bottleneck almost perfectly. A good philosophy undergrad has spent four years being told that their argument has a hidden assumption on page three, that their definition of "consent" doesn't cover the edge case, that their thought experiment breaks under a particular substitution. That is eval design. That is red-teaming. That is prompt engineering with the serial numbers filed off.
CS programs, meanwhile, have spent the last decade optimizing for a job market that no longer exists in the same shape. The median CS graduate in 2026 has been trained to solve a well-specified problem quickly; the market now pays for the ability to specify the problem in the first place. The delta between those two skills is the delta between a $95k implementation role and a $240k "forward-deployed" role, and it is widening.
The community reaction on Hacker News is instructive. The top comment, at 340 points, is a CS grad from 2023 pointing out that his philosophy-major roommate makes twice what he does at an AI lab and "literally just writes documents all day." The second-highest reply is a hiring manager at a mid-sized startup saying they stopped filtering resumes by degree eight months ago and their hire quality went up. The pushback — that this is a temporary market anomaly, that the philosophy majors will get automated next — is real but underweighted. If the model gets better at specification, the humans who are good at specification get more leveraged, not less.
If you are a working developer reading this, the actionable question is not "should I have majored in philosophy." It is: what portion of your daily output is implementation, and what portion is specification, judgment, and taste?
If the answer is that you spend most of your day translating tickets into code, you are competing with an English major who has a Claude subscription and thirty hours of practice. That is not a comfortable position. The people who are pulling ahead in this cycle are the ones who have quietly repositioned toward the parts of the job that models are worst at: deciding what the product should do, catching the subtle wrongness in a generated diff, writing the eval that proves the feature actually works under adversarial input, and holding the mental model of the whole system tightly enough to notice when a change three modules away breaks an implicit invariant.
Concretely: get good at writing. Not blog-post writing — specification writing. Write your tickets like a philosophy paper: define your terms, state your assumptions, enumerate the edge cases, name the acceptance criteria in a way that a model could grade itself against. The developers I've seen thrive in the last eighteen months are the ones whose PR descriptions read like proofs and whose eval suites read like exam questions. That skill compounds. Grinding a fifteenth medium on LeetCode does not.
The second move is to stop treating the model as a code generator and start treating it as a very fast, very literal junior engineer who will do exactly what you say and nothing else. That reframe forces you to write better specifications, which is the same skill the philosophy majors are getting paid for.
The most likely 2027 outcome is not that CS degrees become worthless — the people who can do both the specification and the systems-level implementation will remain the highest-paid engineers in the market. The likely outcome is that the floor of the CS job market keeps eroding while the ceiling on "can articulate a hard problem clearly" keeps rising, and the two curves meet somewhere uncomfortable for the median implementation-focused developer. The philosophy majors aren't the threat. The threat is that the job stopped being about the code and nobody sent a memo.
This article seems high on vibes, low on metrics.> While a plain-vanilla philosophy degree remains as hard to monetize as ever, David Chalmers, a prominent philosopher of consciousness at N.Y.U., observes: “I think the demand for philosophers with A.I. training is, if anything, outstripping the s
I studied analytic philosophy, which is basically an education in how to clarify your thoughts, say what you mean in precise terms, and make clear arguments. IMO there is no better preparation for any sort of writing-and-thinking job than studying analytic philosophy, although of course I am biased.
I've spent a surprising amount of time reading philosophy of language, and it's probably done more for my AI prompting than most of the "prompt engineering" articles I've read.Speech Act Theory, Austin's How to Do Things with Words, and Searle's work changed how I
In summary, AI has tricked a bunch of philosophy majors into not only thinking it's more than linear algebra but changing their entire life trajectories because of their confusion. AI seems to be a very alluring tar pit for the non-technical. The sad part is how this negative externality of AI
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
I have an undergrad in Philosophy. Our required math classes were formal logic, which we shared with the comp sci. majors. Those classes basically taught me how to program.I am now a senior engineer at a F500 and I just have a BA in Phil.It is a highly underrated degree, and going forward with knowl