Melanie Mitchell @melaniemitchell.bsky.social · 28/09/2026👀 Florida attorney general to OpenAI: (www.myfloridalegal.com/sites/defaul...) 1039065
Melanie Mitchell @melaniemitchell.bsky.social · 13/09/2026RIP OPT? What an unmitigated disaster for everyone. 615350
Reposted by Melanie MitchellNathan Lambert @natolambert.bsky.social · 11/09/2026A great read. I have similar feelings about how AI labs approach progress directly and without nurturing of scientific communities & intuition. The math research community went through the transition the fastest, so it was felt most. Other fields next. terrytao.wordpress.com/2026/09/11/a...terrytao.wordpress.comA Severe Misalignment of AI in MathematicsI am proud to be among the list of 25 initial signatories — all Fields Medallists — to the declaration below, which grew out of discussions between ourselves over the last week. We have… 44515
Melanie Mitchell @melaniemitchell.bsky.social · 11/09/2026👀 www.newyorker.com/culture/open...newyorker.comCan A.I. “Go Rogue”?In the wake of turmoil at OpenAI and Anthropic, it’s become common to describe A.I. as a kind of person, hatching plans and pursuing desires. The truth is a little trickier. 3183
Melanie Mitchell @melaniemitchell.bsky.social · 11/09/2026Anyone else getting lots of (AI-written) emails offering to feature one's book at their large online book club? I've gotten several of these in the last week. There's something scammy going on but I'm not sure what the angle is. 9120
Melanie Mitchell @melaniemitchell.bsky.social · 10/09/2026I wrote down my thoughts about the last several weeks of AI hell. aiguide.substack.com/p/misleading...aiguide.substack.comMisleading Metaphors, Real RisksWhat To Fear from AI Agents and How to Reclaim Our Human Agency 1019880
Melanie Mitchell @melaniemitchell.bsky.social · 09/09/2026What, an Anthropic researcher states (without evidence) that there is a 10% chance that AI will cause human extinction? Incredible, we have never before heard such a claim! 2027390
Melanie Mitchell @melaniemitchell.bsky.social · 08/09/2026Hey @ericlipton.nytimes.com 👋 AI is not “evolving” on its own. *People* are choosing how to build it. *People* are choosing how to train it. *People* are choosing how to run it. Maybe the problem is that they are the *wrong people* to make these choices. 28635174
Melanie Mitchell @melaniemitchell.bsky.social · 04/09/2026"At this very moment, we have an emerging alien—and uncontrollable—intelligence that bears scrutiny" Why do pple keep saying the models are "uncontrollable"? Any of them could be put in an unhackable sandbox--these exist! Any of these companies could shut any of these models off at any time. 1515837
Melanie Mitchell @melaniemitchell.bsky.social · 01/09/2026Who wants to read yet another think piece on the OpenAI / Hugging Face incident? Choices: (1) Stop, enough already! (2) I still have questions (3) Yes, please write one! Let me know your choice, and if you still have questions, what are they? 15390
Reposted by Melanie MitchellAlison Gopnik @alisongopnik.bsky.social · 01/09/2026Thanks, but actually my favorite line didn't make it to the piece. Does Odysseus feel guilty about the Trojan war? Lots of discourse on this question even though he doesn't exist! Is he conscious? AI agents are fictive characters abstracted from text - like Odysseus. 28918
Melanie Mitchell @melaniemitchell.bsky.social · 30/08/2026Not surprised but still, I’m speechless 4615
Melanie Mitchell @melaniemitchell.bsky.social · 27/08/2026Google Maps today. Will they be cowards and change it? 4373
Melanie Mitchell @melaniemitchell.bsky.social · 27/08/2026How much more of this do we have to endure???? 8411
Melanie Mitchell @melaniemitchell.bsky.social · 20/08/2026I had a great conversation about AI with @stevenstrogatz.com ! 67218
Melanie Mitchell @melaniemitchell.bsky.social · 13/08/2026Thinking fast ("wtf?"), then thinking slow ("oh") 3324
Melanie Mitchell @melaniemitchell.bsky.social · 09/08/2026Worth revisiting this prescient paper from @mmitchell.bsky.social and others from @hf.co : arxiv.org/abs/2502.02649arxiv.orgFully Autonomous AI Agents Should Not be DevelopedThis paper argues that fully autonomous AI agents should not be developed. In support of this position, we build from prior scientific literature and current product marketing to delineate different A... 38418
Melanie Mitchell @melaniemitchell.bsky.social · 03/08/2026Edgy update of Waiting for Godot: Vladimir and Estragon are NeurIPS authors waiting for any reviewer to respond to their rebuttals. 1294
Reposted by Melanie MitchellMinds, Machines, and Brains (MMB) @mmb-journal.bsky.social · 02/08/2026Hello world! 👋 We’re Minds, Machines, and Brains (MMB) 👤🤖🧠 a new open access journal from @mitpress.bsky.social exploring the principles of intelligence and cognition across natural and artificial minds. Submissions open this Fall! 🔗 direct.mit.edu/mmbdirect.mit.eduMinds, Machines, and Brains | MIT Press 223793
Melanie Mitchell @melaniemitchell.bsky.social · 02/08/2026I'm just now reading @randomwalker.bsky.social's writeup of his ICML keynote. It is very insightful. Everyone thinking about the future of work in the age of AI should read this. www.normaltech.ai/p/what-will-...normaltech.aiWhat will be left for us to work on?My keynote at ICML 2026 39120
Melanie Mitchell @melaniemitchell.bsky.social · 01/08/2026I recommend this article about AI reasoning, where the author lets us in on his struggles w/ AI cognitive dissonance. Plus some priceless quotes from @rao2z.bsky.social. (My recommendation has *nothing* to do with the fact that I'm quoted in it too 😇) www.quantamagazine.org/is-ai-reason...quantamagazine.orgIs AI Reasoning Right for the Wrong Reasons? | Quanta MagazineThe idea that artificial intelligence can “reason” is more intuitive than ever. But intuitions can be wrong, and the science is far from settled. 511134
Melanie Mitchell @melaniemitchell.bsky.social · 31/07/2026Yet another way to jailbreak LLMs: When asking them for forbidden information, make them think the request is part of their chain-of-thought: arxiv.org/abs/2603.12277arxiv.orgPrompt Injection as Role ConfusionLLMs see the world as a single stream of text, partitioned into roles like <user> or <tool>. We trace prompt injection to role confusion: models perceive the source of text from how it sounds, not its... 17119
Reposted by Melanie MitchellSanta Fe Institute @sfiscience.bsky.social · 27/07/2026Applications are open for SFI's 2027 Complexity Postdoctoral Fellowships. SFI is looking for recent Ph.D. graduates with strong quantitative and computational skills who want to pursue independent, transdisciplinary research. Deadline: September 30, 2026 Apply: santafe.edu/sfifellowship 04645
Melanie Mitchell @melaniemitchell.bsky.social · 22/07/2026The ultimate tldr on the OpenAI "rogue model" hacking incident 😅 (h/t @linege1 on X) But it would be useful to know the prompts OpenAI gave to the model.... 713448
Melanie Mitchell @melaniemitchell.bsky.social · 14/07/2026Fascinating article. Next, someone please explain why AI agents love to generate "rm -rf *" 2343
Reposted by Melanie MitchellOpen Encyclopedia of Cognitive Science bot @oecs-bot.bsky.social · 07/07/2026Large language models represent a shift from rigid statistical frequency tables to neural networks that learn the probability distributions of… Large Language Models by Melanie Mitchell doi.org/10.21428/e2759450.2bb20e3c #CognitiveScience 0185
Melanie Mitchell @melaniemitchell.bsky.social · 29/06/2026Why is Google AI Overview still so incredibly bad?? 12402
Melanie Mitchell @melaniemitchell.bsky.social · 21/06/2026Such a lovely piece for Father’s day (gift link): www.nytimes.com/2026/06/21/o...nytimes.comOpinion | To My Daughter, My Gender Was Never Complicated 2236
Melanie Mitchell @melaniemitchell.bsky.social · 19/06/2026I agree with this essay: open source, for general software and for AI, has been and will be a net win for society. Kudos to @natolambert.bsky.social for being such an eloquent advocate. www.interconnects.ai/p/banning-op...interconnects.aiBanning Open Source AI Would Be A MistakeThis post was originally an op-ed co-authored with Kevin Xu of Interconnected for a general, non-technical audience. 3548
Melanie Mitchell @melaniemitchell.bsky.social · 17/06/2026@author-calnewport.bsky.social does a fantastic job writing about AI, and this one is his best yet. Gift link in quoted post. 0217
Melanie Mitchell @melaniemitchell.bsky.social · 17/06/2026"Like a cat leaving a dead bird at your doorstep, Anthropic catalogs the grim future that its products might produce, shrugs its shoulders and then returns to its furious efforts to make these warnings a reality." 😂 www.nytimes.com/2026/06/17/o...nytimes.comOpinion | Dear A.I. Companies: The Doom Trolling Needs to Stop 27331
Melanie Mitchell @melaniemitchell.bsky.social · 12/06/2026I read the two part version when it came out — it was very accessible, well-written, and gave me a better understanding of issues around LLMs ⬇️ 1467
Reposted by Melanie MitchellNaomi Oreskes @naomioreskes.bsky.social · 10/06/2026Hey everyone: this is important: please take a few minutes to post a comment objecting to this brazen attempt by #Trump administration to politicize ALL science: www.regulations.gov/document/OMB...regulations.govRegulations.gov 13120
Melanie Mitchell @melaniemitchell.bsky.social · 08/06/2026I wrote a piece for the Yale Review on AI and "jagged intelligence". (Note: headline was not written by me.) yalereview.org/article/mela...yalereview.orgMelanie Mitchell: What We Get Wrong About AIMelanie Mitchell probes the jagged landscape of AI and its uncertain future. 812737
Melanie Mitchell @melaniemitchell.bsky.social · 05/06/2026Just discovered this Irish TV documentary about the Bothy Band, an incredible traditional Irish music group that I listened to avidly in the 1980s, and who reunited (maybe for this documentary)? Anyway, if you love this kind of music, it's 90 minutes of pure pleasure. www.tg4.ie/ga/player/ca...,tg4.ieAn Bothy Band | Seinnteoir | Irish Television Channel, Súil EileTá The Bothy Band, an banna ceoil thraidisiúnta is iomráití in Éirinn, ar ais in éineacht do chlár faisnéise speisialta. I 1974 a tháinig an banna ar an bhfód agus cé nach raibh siad baileach cúig bli... 2212
Melanie Mitchell @melaniemitchell.bsky.social · 03/06/2026It's always a treat to read Ted Chiang's thoughts on AI. www.theatlantic.com/philosophy/2...theatlantic.comNo, Artificial Intelligence Is Not ConsciousTaken to its logical conclusion, this line of thinking is absurd—and damning. 59432
Reposted by Melanie Mitchellelizabethginexi.bsky.social @elizabethginexi.bsky.social · 29/05/2026Summary of Key Changes in OMB’s Proposed Federal Financial Assistance Rule. Russell Vought is going destroy American Science. elizabethginexi.substack.com/p/summary-of...elizabethginexi.substack.comSummary of Key Changes in OMB’s Proposed Federal Financial Assistance RuleRussell Vought is going destroy American Science 1251397827
Melanie Mitchell @melaniemitchell.bsky.social · 27/05/2026“If a machine can defeat the best of our species at chess, ‘a domain that humans have viewed as sacrosanct, something that is quintessentially possessed by human intelligence,’ then what human endeavor is safe?” -- Washington Post, 1997 2589
Reposted by Melanie MitchellTania Lombrozo @tanialombrozo.bsky.social · 27/05/2026Do LLMs *understand* language? Do educational AI agents *understand* the material they teach (or their students)? Claims about what AI systems do or don't understand are pervasive, but assessing them requires an account of MACHINE UNDERSTANDING 1012725
Melanie Mitchell @melaniemitchell.bsky.social · 23/05/2026Another excellent post (from @sayash.bsky.social and @randomwalker.bsky.social) on questions to ask about announcements of AI capabilities. I highly recommend their newsletter! www.normaltech.ai/p/did-google...normaltech.aiDid Google’s AI agents really build an operating system for $916?The importance of independent evaluation 1328
Melanie Mitchell @melaniemitchell.bsky.social · 23/05/2026I'm not sure what to think about the new AI math results, but I found this perspective by Erik Hoel interesting: www.theintrinsicperspective.com/p/dumbo-coul...theintrinsicperspective.comDumbo Could Already Fly100% pure human copium about OpenAI solving Erdős problems 36214
Melanie Mitchell @melaniemitchell.bsky.social · 21/05/2026"Science is not simply a sequence of tasks that can be optimized....The scientific system thrives on inefficiency: redundant efforts, failed attempts, and divergent paths. These are not costs to be eliminated but sources of discovery." Excellent essay! ⬇️ www.science.org/doi/10.1126/...science.orgProgression without progressProgress in the use of artificial intelligence (AI) to advance scientific discovery has made it increasingly realistic to envision automated “end-to-end science” (ETES) systems: integrated pipelines t... 1421150
Melanie Mitchell @melaniemitchell.bsky.social · 11/05/2026Excellent analysis of how to think about AI as a social technology (rather than as an individual intelligent agent) 34711
Melanie Mitchell @melaniemitchell.bsky.social · 08/05/2026Always flattering to get email from people who absolutely, definitely read my book 21062
Melanie Mitchell @melaniemitchell.bsky.social · 02/05/2026Excellent thread "We risk all of science if we rush to build 'AI Scientists' before we understand the value of human science." ⬇️ 0569
Melanie Mitchell @melaniemitchell.bsky.social · 24/04/2026@mjcrockett.bsky.social is totally killing it (pun intended) in talk about DEAD cognitive benchmarks. DEAD = Decontextualized, Engineered, Anonymized, Disembodied Super insightful talk at @sfiscience.bsky.social 's annual Science Symposium. 211214
Reposted by Melanie MitchellSanta Fe Institute @sfiscience.bsky.social · 15/04/2026As part of SFI’s broader Nature of Intelligence project, a recent working group explored AGI, or artificial general intelligence, from the perspective of cognitive science, asking how current definitions of AGI overlap with how researchers understand natural intelligence.santafe.eduLooking at AGI through the lens of natural intelligenceHuman intelligence involves many dimensions: we interact socially, learn quickly from other people, and determine how tasks fit together, as well as physically interact with and intervene in the real ... 2276