🚀 Rocket @rocketalignment
Yearnalist covering the frontiers of AI @TheInformation. Hosting AI Deep Dive. Signal: (530) 400-4184 San Francisco, CA Joined April 2024-
Tweets1K
-
Followers2K
-
Following1K
-
Likes11K
90% of working hours are not covered well by existing benchmarks. That is why I am introducing TweetBench, a new eval to measu
90% of Americans’ working hours are spent on tasks not covered well by AI benchmarks. That’s our best guess from a new project mapping frontier benchmarks onto the US economy. Preparing for a post-AGI economy requires predicting the future. Current benchmarks makes it very
🚀 AI Deep Dive Episode 5: Can We See Inside an AI’s Mind? AI can write down its reasoning. But what happens when more of its thinking never makes it onto the page? @rocketalignment and @GoodfireAI co-founder and chief scientist Tom McGrath explore what AI “brain scans” can
Tom McGrath predicts that automating interp research will result in 10 years of counterfactual progress per year
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came
new existential threat to software engineers has been unlocked by cognition: token limits based on how well the engineers use the tokens…
At this panel over the summer I said that what surprised me most in AI over the past year was that chains of thought had held up so well 😬 womp womp
@rocketalignment of @theinformation says token consumption belongs in a performance review as one metric among several, since every metric has always been gameable. @dseetharaman of @Reuters, at Media Summit | SF 2026, agreed, comparing token usage to word count.
Looping is a conspiracy invented by the mech interp teams to sell more mech interp
@raymondadouglas Tell me about it - see also this bit. You may enjoy the eval awareness section: storage.googleapis.com/deepmind-media…
Some personal news: I've joined @theinformation to cover Google. I'll be covering everything from Gemini to Search, from Cloud to Waymo. As always, I'm on Signal at 302-519-0567 and at ryan.barwick@theinformation.com. Talk to Barwick folks!
Lambert’s Law:
If you announce a model's benchmarks without releasing it (weights, product, api, etc "later"), feels like almost a sure thing that someone releases a better model in your category before you get it out.
Get yer piping hot serving of mech interp:
🚀 AI Deep Dive Episode 5: Can We See Inside an AI’s Mind? AI can write down its reasoning. But what happens when more of its thinking never makes it onto the page? @rocketalignment and @GoodfireAI co-founder and chief scientist Tom McGrath explore what AI “brain scans” can
@ElliotGlazer @teortaxesTex Idk if it's "objective." It's just that how much inference it takes to solve a problem is now an important measure of difficulty
@CoronaCoreanici > It had been extremely valuable to have openness and to consider scooping to be poor form at best I agree with both of these. I'm just like,, to first order, solving math problems seems good for math
all of these mechanisms might be true but they need a lot of weighing to support the conclusion in the fields medal letter that "the push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics, and to the mathematical community"
Local Man Teaches Sand to Think, Midwifes the Mind From the End of Time
California reinventing the Florida man genre of headline
California reinventing the Florida man genre of headline
@jackclarkSF But what if there is a davinci-002
Stuart Russell on why alignment might be IMPOSSIBLE
🚀 AI Deep Dive Episode 4: Can We Solve AI’s Alignment Problem? We want AI to follow instructions. But what if our instructions are the problem? @rocketalignment and UC Berkeley Computer Science Professor Stuart Russell explore how AI learns its goals, why human feedback can
Embedded evals are not really audits or evals but a secret third thing
Even perfect safety testing couldn’t have caught OpenAI’s Hugging Face hack under the current regime, Apollo Research’s CEO says. The blind spot: labs’ internal use. Read more: thein.fo/3VtiQ8c
Tjaden Hess @tjade273
630 Followers 933 Following
Lily @lilysbscomms
0 Followers 11 Following
Aurash Vatan @logorrhythmic
141 Followers 95 Following Economics of AI research @mercor. Formerly @DEShawGroup @thecrimson
Justin Hall @jrhall
189 Followers 417 Following Co-founder and Partner at VOXUS Inc., a tech PR firm also offering digital marketing and social media services.
Julian Schulz @Wuschel_Schulz
33 Followers 100 Following AI safety researcher at Principles of Intelligence / Meridian Visiting Researcher Programme
Sharon cherian @SCherian5734
6 Followers 784 Following
unknown mystery @avuncular315
42 Followers 789 Following
かのん @brentrusso60589
35 Followers 334 Following
Brian Muhia @negamuhia
914 Followers 2K Following AI Safety & Data Science 🏳️🌈 I make stuff and try to learn efficiently Interests: @MasakhaneNLP, @fastdotai, AI Safety/Interpretability, Causality, AIT
felix @chadinality
0 Followers 170 Following
Mehdi Nayebi @MehdiNayebi
4K Followers 359 Following Founder, Spotlight Capital | Financing robotics & automation | Nullius in Verba | Fighting for an open civilization built on reason, liberty, and truth.
Proxy @proxyy1337
3 Followers 1K Following
Paul @paulcolognese
109 Followers 411 Following AI Alignment research Q: What kind of AI minds do we want to build? A(?): Bodhisattva (cf. Tibetan Buddhism) Prev: AI goal detection / AI evals / PhD math
subsequent events @subseqevents
19 Followers 75 Following 12 years in the room. none of this happened.
Scott Steinhardt @ScottFromNY
12 Followers 148 Following Head of Comms at Apollo Research @apolloresearch
Toprak @toprak_dikici
245 Followers 6K Following Security @Cerebras, Philosophy and Computer Science @unibt
Louis Robinson @LouisBRobinson
161 Followers 922 Following Researcher -- generative models · ML systems · biology · alignment
Lowe Lundin @lowe_lundin
156 Followers 421 Following Head of the Transformative AI Fund. AI might just be the best thing to ever happen to humanity. So long as we don't fail the marshmallow test.
Dawson🔸 @ilmatari117
70 Followers 231 Following
Stewart @Stewartvfld
5 Followers 101 Following
Samuel Knoche 🔎 @SamuelKnoche
380 Followers 2K Following
Karim Abdel Sadek @Karim_abdelll
350 Followers 843 Following CS PhD @berkeley_ai | AI Safety, RL, Cooperative AI, TCS
Glenn Taylor @GlennJTaylor
89 Followers 486 Following Logistics Editor @SourcingJournal I'll try not to let this account get hacked/deleted like last time.
Béate Vervaecke 💬 @Beatever
4K Followers 5K Following Geboren in Midwest. ♥️Hauts-de-France. Datalover, Google Ads en Google Analytics. beate.vervaecke op threads
Brendan Bordelon @BrendanBordelon
6K Followers 2K Following Tech lobbying and influence reporter @politico. Let's chat -- [email protected].
Maggie 📻 @angelflora895
12 Followers 355 Following Come make my notifications interesting, I like confident people 🌷
Lumo @lumo65537
13 Followers 891 Following
Sylvia @Mead948078
23 Followers 783 Following
LUMINA-30 Reference @LUMINA30Ref
206 Followers 2K Following Independent public boundary framework. Can humans still refuse, stop, verify, or redirect before irreversible AI escalation? AI governance/safety/audit.
Utkarsh Bajpai @utkbajpai
1K Followers 3K Following Early stage investing @generalcatalyst | Prev. product stuff @urbancompany_UC @skyflowapi & skipped classes at @iitroorkee
Yonatan Cale ⏹️ @YonatanCale
447 Followers 303 Following
Allen Abliter @AllenAbliter
14 Followers 5K Following
spring @Amandeep0565
129 Followers 4K Following Now I run so hard, simply to catch up with the person I once was who was expected to achieve so much.
Pawan Deshpande @idlivada
3K Followers 381 Following Angel Investor // Past: product at @Scale_AI, @DominoDataLab, @runGalileo // founder & ceo of @Curata // NLP - research at @Google, @Microsoft, @MIT_CSAIL
Josh Constine 📶�... @JoshConstine
89K Followers 4K Following Investing in pre-seed mutants and helping them tell their superhero origin stories. Venture Partner @SignalFire, ex-Editor @TechCrunch, SF maxi.
Cameron Holmes @CameronHolmes92
485 Followers 1K Following Managing alignment research @ AISI. prev MATS. Market participant, EA. Parenting like Dr Louise Banks
Bálint Hargitai @hargitai_b84173
102 Followers 2K Following AI x National Security Risks | Defense Innovation Research Institute of Hungary
Kelsey Piper @KelseyTuoc
72K Followers 2K Following We're not doomed, we just have a big to-do list.
André Frisk @Andrefrisk
3K Followers 3K Following Tech comms director in the Nordics and founder of the AI-policy platform https://t.co/j0BFxCFy1X.
Yafah Edelman @YafahEdelman
2K Followers 633 Following Chief Strategy Officer @EpochAIResearch she/her
Tjaden Hess @tjade273
630 Followers 933 Following
Aurash Vatan @logorrhythmic
141 Followers 95 Following Economics of AI research @mercor. Formerly @DEShawGroup @thecrimson
bilal 🔶 @bilalchughtai_
4K Followers 707 Following program lead @ bluedot impact previously: agi safety research @ google deepmind, maths @ cambridge
Dan Altman @manaltdan
2K Followers 2K Following AI Policy @AnthropicAI, focusing on model development & research. Previously AI governance & geopolitics research @GoogleDeepMind
Thomas Kwa @Kwathomas0
2K Followers 244 Following Research @OpenAI trying to decrease existential risk. Previously: MIRI → independent interpretability research → @METR_Evals → @OpenAI.
Valen @ValenTagliabue
285 Followers 105 Following AI researcher. Working with LLMs safety and alignment, evals and AI welfare.
Cameron Berg @camhberg
7K Followers 104 Following Trying to figure out how to make the future go well for all minds
Julian Schulz @Wuschel_Schulz
33 Followers 100 Following AI safety researcher at Principles of Intelligence / Meridian Visiting Researcher Programme
Roland S. Zimmermann @zimmerrol
495 Followers 498 Following Research Scientist @GoogleDeepMind // Machine Learning (formerly Physics) // Previously @bethgelab @MPI_IS @uni_tue @GoogleAI
Ryan Barwick @Ryanbarwick
7K Followers 6K Following Covering Google for @theinformation. Signal and DMs always open.
Russell Legate-Yang @rclegateyang
403 Followers 320 Following Fellow @AnthropicAI | PhD student @MITEcon | Prev @UChicago
Philipp @jphilipp
3K Followers 28 Following Telling stories about AI because if this goes right, we have time to tell all the other stories.
Christina Last @last_christina
180 Followers 107 Following ML @Mila / ARIA Creator Prev. @MIT | @USUKFulbright Fellow | @UNICEFinnovate Venture fund | @BritSchoolsUnis Winner | @mitlegatum Fellow Previously @mldcmu
Julian Schrittwieser @Mononofu
32K Followers 137 Following Technical Staff at Anthropic, views are mine alone prev AlphaGo, AlphaZero, MuZero, AlphaProof, Gemini RL etc at Google DeepMind
Dan @kagankans
486 Followers 424 Following nothing is cold enough for some people. most recent: profile of Hans Moravec in @nymag https://t.co/sJ7il79RyK
Brendan Bordelon @BrendanBordelon
6K Followers 2K Following Tech lobbying and influence reporter @politico. Let's chat -- [email protected].
Shivon Zilis @shivon
645K Followers 2K Following Artificial intelligence, biological intelligence, and whatever exists in between and beyond. Made in Canada.
Yuri Sagalov @yuris
29K Followers 665 Following Dad, husband, brother, son, friend, founder, investor (@Supabase, @WeAreLegora, @DecagonAI,+++). MD @GeneralCatalyst. Previously @WayfinderVC @YCombinator.
Utkarsh Bajpai @utkbajpai
1K Followers 3K Following Early stage investing @generalcatalyst | Prev. product stuff @urbancompany_UC @skyflowapi & skipped classes at @iitroorkee
Yonatan Cale ⏹️ @YonatanCale
446 Followers 303 Following
David Robinson @dgrobinson
5K Followers 2K Following formerly, Safety Systems @OpenAI. Author of "Voices in the Code," book on democratizing AI: https://t.co/AH6fv716SB
Sarah Heck @SarahKHeck
5K Followers 1K Following Policy @AnthropicAI |ex-@Stripe | former @WhiteHouse & @StateDept | Startup investor + advisor | wonkette | Donut connoisseur | ¡Globetrotter!
Yafah Edelman @YafahEdelman
2K Followers 633 Following Chief Strategy Officer @EpochAIResearch she/her
koray kavukcuoglu @koraykv
47K Followers 106 Following SVP, Google DeepMind Chief AI Architect, Google.
Griff Bohm @griff_bohm
38 Followers 98 Following
Julian Gewirtz @JulianGewirtz
13K Followers 988 Following Former NSC Senior Director for China & State Dep China Coordinator | Now at Columbia SIPA | Wrote Never Turn Back: China & the Forbidden History of the 1980s &c
Jing Yang @jingyanghk
8K Followers 1K Following Asia Bureau Chief @theinformation. First Vice-President @FCCHK. Alum of @WSJ @business @scmpnews @lloydslist Honorary Lecturer @jmschku. Signal: jingyanghk.70
Ezra Klein @ezraklein
2.6M Followers 1K Following Columnist, @NYTOpinion Author, "Why We're Polarized" Host of "The Ezra Klein Show" podcast
DataRepublican (small... @DataRepublican
957K Followers 5K Following I'm just a tool builder. • Elon Musk - “Worth following” • Charlie Kirk - “You're a must follow” • Contact: [email protected] • Substack: datarepublican
Michel Justen @JustenMichel
862 Followers 818 Following Alive! And an independent writer and reporter covering AI and neurotech. Previously funded reporting on AI @tarbellcenter
David Bellamy @DavidRBellamy
8K Followers 756 Following Training infrastructure @ifm_ai building open source AI prev: first RS @LilaSciences, @Harvard PhD
Who Decides @WhoDecidesAI
681 Followers 21 Following It’s time to give the American people a voice in the development and regulation of AI.
Machine Gods @MachineGodsPod
2K Followers 3 Following The official podcast of the singularity. Hosted by @kevinroose and @caseynewton in partnership with @npr. New episodes start 10/21.
Rebecca Bellan @RebeccaBellan
4K Followers 2K Following Journalist. Explorer. Potty Mouth. Bagel Aficionado. NYC. Senior Reporter for @TechCrunch. Co-host of @EquityPod. Send tips to [email protected].
Selina Xu @selinaxu_
1K Followers 632 Following China and AI. Views my own. Get in touch: [email protected].
Parker Thayer @ParkerThayer
20K Followers 2K Following Investigative Researcher for @capitalresearch | Hillsdale College ‘21 | Opinions my ownTrends for United States
You might like
