Robert O’Callahan resigned from Google on September 24 after years there, most recently building tools for chip design. His reason, stated plainly in the note he sent colleagues: those tools make AI cheaper and faster, and he thinks AI is moving faster than people can adapt to. He could not find a Google project that would not accelerate AI in some way, so he left. The 251-comment thread on Hacker News is part sympathy, part argument about whether that position is coherent.

What he says

  • He did not work on AI capability itself. He worked on hardware chip design tools, told himself for a while the work was “relatively harmless,” then concluded the main impact would be accelerating a new breed of AI chips. He told his skip manager he was “at best reluctant to see their project succeed.”
  • Staying wasn’t an option he could engineer. He lives in New Zealand, Google has no other engineering teams there, and he says he tried to have influence from inside Google DeepMind — that effect “does not seem to be strong.”
  • The motive is religious as well as technical: “It’s tempting to just turn a blind eye to the impact of my work, but that would not be a Jesus-following thing to do.”
  • He is neither a doomer nor a dismisser. Phenomena the doomers predicted have arrived, by his account — reward hacking, deceptive models, “model eval awareness” — but he won’t put catastrophic risk at 100%: “I think the risk is real but uncertain — but that itself is very alarming!” Aiming for superintelligence in the near future is, in his words, “inherently irresponsible.”
  • He explicitly rejects the “AI is a scam” camp: “Many prominent AI detractors (looking at you, Zitron and Doctorow) seem to think that AI is some kind of scam that won’t really work. I think it will.” His objection is to the rate, not the technology’s viability.
  • On the labs’ sincerity, he says he has talked to many people at Google DeepMind and found “almost all of them have sincere and serious concerns,” and that he has seen no hard evidence AI risk is being hyped to pump share prices.
  • His sharpest structural point is about time. PCs, the internet and smartphones left large areas of human activity untouched; this doesn’t. So the ordinary instruments people use to plan a life — a degree, a career, a few years of training — stop working, because they assume the world will hold still long enough to finish.

What the thread adds

  • thinkingemote — the clarification most of the thread needed, since the note reads as a rejection of AI it doesn’t quite make: “He’s not even leaving ‘because AI’. He’s literally leaving to focus more on AI stuff.” They quote his plans back at him — studying how AI agents debug code, using agents on his own projects — and land on “The main reason then is the concern of the velocity of the rollout and the increased pervasiveness of AI. Too fast, too much.”
  • roca — the author, in the thread, confirming a detail the audience couldn’t know. teiferer observed the note reads less like something you send HR and more like something you send your soon-to-be-ex-colleagues; roca replied: “Indeed, that message was for my coworkers. HR gets a form.”
  • negura — what was new to them: the idea that society’s planning horizons are calibrated to a rate of change. A four-year degree, a five-year career bet, a mortgage — all assume the world shifts at roughly the speed it always has. “If the rate of disruption exceedes that assumption, the timeframes become inadequate.”
  • glimshe — the genre history. The “Why I left Google” post used to be about free food and minor gripes; now, at OpenAI, Anthropic and Google alike, “it’s ‘I had to leave because I could have destroyed humanity!’”
  • NiloCK and pu_pe — the same question from two directions. pu_pe: “We had people departing from OpenAI, Anthropic and now Google… At which point do we believe the message?” NiloCK’s correction: this isn’t new, Hinton left Google in 2023 from a much more influential perch for similar reasons — “it’s just newly breaking through into mainstream conversation.”
  • Gareth321 and cromka — the answer to heinternets’ “Is it consuming too much science fiction that has people thinking like this?” Gareth321 describes having considered the worry histrionic a year ago and changing their mind after using the tools, then points to Hinton’s public talks; cromka is blunter: “The guy was an insider for years but you call him a noob who just read a fantasy book?”
  • jdub — the upside nobody else raised: engineers this senior leaving the labs may mean “the liberation of creative minds. So many fantastic open source contributors have disappeared into corporate black holes.”

Where the thread pushes back

  • youoy argues the leverage is inside, not outside: “One person quitting is nothing for the company. The risk of big group quitting at the same time on the other hand… The labor is the only thing the company cares about. So we cannot wait until we are all replaceable by AI and we lose that leverage.” jhanschoo answers that not everyone can “steer the ship” from within, and that the people visibly engaging with the problem are the ones leaving.
  • injidup aims at the theology rather than the engineering: why is uncertainty about one form of superintelligence terrifying while uncertainty about an omnipotent one is comforting? It drew ten replies, including fwlr (“God’s been around for thousands of years and we are still here and flourishing, so obviously God is in some way ‘aligned’; the same cannot be said for AGI”) and Joeri on the free-will tension.
  • A scepticism cluster — keithnz (“struggling to see an actual reason other than he doesn’t like the pace”), ambivalence (“oh no, AI too powerful too quick. let me quit to work on tools to make AI better at debugging code. excellent logic, m’lord”) and xnx (“Probably my jealousy, but it’s hard not to read these as ‘I am rich.’”).
  • grahar64 and Gareth321 end up on the governance question underneath it: if you thought humans were a net positive you wouldn’t spend trillions replacing them — with Gareth321’s counter that the real failure is governance and distribution, not the machines.

The question the thread kept asking

Several commenters arrive at the same dead end from different doors: what does believing the warning actually license? pu_pe asks at what point belief is warranted, BLKNSLVR notes the asymmetry that people leave over AI but almost never over advertising (“The more people leaving ‘because AI’ just makes me sad that there have been so few leaving ‘because advertising’”), and ink states the fatalist version directly — companies have no incentive to stop, states can’t risk falling behind, “There is nothing we can do anymore.” O’Callahan’s post does not answer it either; he says only that he wants his future work to be “unambiguously pro-human.”

Comments are pseudonymous handles, HN publishes no per-comment scores, and the ordering above reflects HN’s own ranking — a slice of the thread, not a consensus.