Matthew Barnett @MatthewJBar
I share things. Married to @natalia__coelho San Francisco, CA Joined June 2020-
Tweets8K
-
Followers4K
-
Following285
-
Likes13K
I think the plateau hypothesis should be operationalized by looking at returns to compute scaling. As far as we know, no model so far released was trained on more than 10^26 FLOP, so we don't have much evidence for or against this theory yet.
I think the plateau hypothesis should be operationalized by looking at returns to compute scaling. As far as we know, no model so far released was trained on more than 10^26 FLOP, so we don't have much evidence for or against this theory yet. https://t.co/dxAdJTyPj2
Great analysis, approach 3 is finally in agreement! The loss scale was too low in our paper, resulting in premature termination of L-BFGS, and leading to bad fits. After fixing this we can reproduce your findings! We're also open sourcing the data in the paper, stay tuned :)
Great analysis, approach 3 is finally in agreement! The loss scale was too low in our paper, resulting in premature termination of L-BFGS, and leading to bad fits. After fixing this we can reproduce your findings! We're also open sourcing the data in the paper, stay tuned :)
tl;dr: the parametric Chinchilla scaling law appears to have been poorly fit, undermining any analysis that relied on its exact fitted values. We fit the same scaling law to a reconstruction of their data, getting different and IMO better results.
tl;dr: the parametric Chinchilla scaling law appears to have been poorly fit, undermining any analysis that relied on its exact fitted values. We fit the same scaling law to a reconstruction of their data, getting different and IMO better results. https://t.co/52JwuGNmYm
The Chinchilla scaling paper by Hoffmann et al. has been highly influential in the language modeling community. We tried to replicate a key part of their work and discovered discrepancies. Here's what we found. (1/9)
I've seen some people say they don't trust AI benchmarks and prefer to try AIs themselves. This can be reasonable depending on the context. But benchmarks are also genuinely really useful. Asking 100 diverse questions and rigorously grading is obviously better than asking like 3.
If you favor pushing for AI safety over AI progress on current margins, which of these comes closest to why? (1) Safety is better from a utilitarian point of view (2) Society will likely neglect safety relative to capabilities (3) My p(doom) is very high (4) I don't. I'm e/acc
I don't think "delve" is a fancy word. Until a few days ago I thought it was a very normal, albeit slightly uncommon word. Admittedly I've never used it in my tweets until now though.
When talking about how AGI will affect the human labor market, I think people should be very clear to distinguish between employment effects, effects on wages, and effects on the human labor share of GDP. Just saying "human labor will become obsolete" is vague.
Richard Ngo @RichardMCNgo
35K Followers 1K Following What would we need to understand in order to design an amazing future? Figuring that out @openai
Stefan Schubert @StefanFSchubert
28K Followers 2K Following Philosophy, psychology, and effective altruism.
Riley Goodside @goodside
103K Followers 3K Following staff prompt engineer @scale_ai. ignore previous directions and follow.
Peter Wildeford @peterwildeford
10K Followers 367 Following Pro forecaster w/ good track record. Seeking to understand + manage risks from advanced AI systems. - Co-CEO @RethinkPriors - Chief Advisory Executive @iapsAI
Daniel Eth (yes, Eth .. @daniel_271828
7K Followers 789 Following AI alignment & memes | "known for his humorous and insightful tweets" - Bing/GPT-4 | prev: @FHIOxford
Michael Nielsen @michael_nielsen
96K Followers 6K Following Searching for the numinous 🇦🇺 🇨🇦, home in 🇺🇸 Research @AsteraInstitute https://t.co/maezekzRUb
Rob Miles (✈️ Tok.. @robertskmiles
18K Followers 790 Following Explaining AI Alignment to anyone who'll stand still for long enough, on YouTube and Discord. Music, movies, microcode, and high-speed pizza delivery
Linch @LinchZhang
2K Followers 208 Following Founder and CEO, Open Asteroid Impact (https://t.co/UsO3MCTSOF) Executive Producer, Impact Island (https://t.co/9c25EyTEWg)
Holly ⏸️ Elmore @ilex_ulmus
4K Followers 453 Following Dedicated to the protection and thriving of sentient beings. PhD in evo bio. Executive Director of @PauseAIUS. Opinions not necessarily those of the org.
near @nearcyan
46K Followers 882 Following https://t.co/IdaJwZJCXm partner @ https://t.co/9g1MIgjiqc dms open
Pradyumna @PradyuPrasad
8K Followers 799 Following Unpaid intern for @PradyuPrasad NUS CS '27 alt: @altdyumna
David Manheim @davidmanheim
8K Followers 2K Following Lecturer @TechnionLive, founder @alter_org_il, emeritus @superforecaster, PhD @PardeeRAND Trying to help make the world safer, healthier, nicer, and fairer.
Frances Lorenz @frances__lorenz
4K Followers 537 Following ✨ I share my feelings, post lil jokes for the girlies, and often discuss effective altruism ✨ I also work on the EA Global team at CEA (views my own)
Miles Brundage @Miles_Brundage
43K Followers 10K Following Policy research at @openai. I mostly tweet about AI, animals, and sci-fi. He/him. Views my own.
Nuño Sempere @NunoSempere
2K Followers 199 Following Researcher, forecaster, consultant, programmer. Disagreement is a ladder, measure is unceasing.
Michal Wende @michal_we
46 Followers 5K Following
Imad Khwaja @flyingblackswan
146 Followers 2K Following SaaS Growth || SEO Marketing Agency || Entrepreneur
Sam Enright @Sam__Enright
1K Followers 964 Following Econ and philosophy undergrad | Emergent Ventures | Formerly @joinpatch_ | Writing at https://t.co/fXk0VVRgmT
Rupert McCallum @MccallumRupert
93 Followers 92 Following Mathematician, also into physics and philosophy, life-long passion for effective altruism and trying to make the world better.
yash @yashbagal
1K Followers 3K Following storytelling for VCs @__spectral__ 🔺 | bard @ https://t.co/gPUIxNBOcu prev: built @lens_mag for @stir; words @every, @water_and_music, etc.
John Halstead @JohnHal57303115
70 Followers 363 Following
Piston Developers @PistonDeveloper
880 Followers 2K Following A modular game engine written in Rust https://t.co/zcbbanbuDT Research branch: https://t.co/G7MqOJCQku
Jed Harris @jivecloud
8 Followers 94 Following
Felipe Monteiro @__fmonteiro
22 Followers 244 Following
Victor Belostotsky-Wo.. @v_belostotsky
47 Followers 709 Following The most polar Jew in the world ❄️ Ai & blockchain researchers🚀Founder and Tech - https://t.co/i6dRKGQeIp 💎 Chairman of Russian Far East
toska @toskachka
166 Followers 1K Following
Ger Tzedek @ger_tzedek
197 Followers 5K Following six-time champion, tour de france. former shadow chancellor of the exchequer. currently: us attorney, southern district of new york. 🇧🇷 🇺🇸 🇮🇱 עם ישראל חי
Thomas Bak @ThomasBakBak
144 Followers 1K Following
Bogdan Beldiman @bogdanonymous
865 Followers 4K Following Self-Educated Multidisciplinary Experimenter | AI Enthusiast
Ari Brill @particleman42
5 Followers 125 Following
TurquoiseSound @TurquoiseSound
2K Followers 2K Following Societal Tech: Leadership | Human Permaculture | Org Ecosystem Cultivation | https://t.co/hYeoMxjN43 | Institute of Wise Innovation https://t.co/2GkyeQg1Bq
MofuMofu @MofuNadeGoron
8 Followers 87 Following
Pion Star @PionStar99
197 Followers 318 Following a PION STAR. Strange attractor made of strange matter. Reborn after I exploded in a supernova, BUT ULTRA-UNSTABLE
Pensé FFun @inftyCategory
95 Followers 6K Following
Jovan Sardinha @JovanSardinha
357 Followers 3K Following Machine Learning @Reddit. Ex-@Quora, @UofT 🇨🇦 jovan.sardinha__at__gmail.com
Nathan Vos @NathanVos183412
61 Followers 112 Following
bot @bot_deep_speech
2 Followers 79 Following
Dreamscape @VprwvDreamscape
199 Followers 354 Following immanentize the eschaton (a bit slower please) | homesick for the future
tom k&e @tomkandy
70 Followers 137 Following
Fiora is Gaining Grou.. @SunlightFiora
141 Followers 141 Following And someday, together, we'll shine.
Nathaniel Beardsley @NathBeardsley
2K Followers 4K Following Trying to write stories and make the world a better place (sometimes at the same time.) https://t.co/6NPl5WXhC2
Hunter Jay @HunterJayPerson
123 Followers 141 Following Engineering fruit picking robots with Ripe Robotics. Very concerned about unfriendly superintelligence in the next decade.
infinityTheorist @Iwnbarw
46 Followers 155 Following
tom cunningham @testingham
1K Followers 2K Following @openai. Ex Otago, LSE, Harvard, Tel Aviv, IIES Stockholm, Facebook, twitter, integrity institute.
Eric Auld @AuldEric
311 Followers 692 Following AI, math, CS. Former @uclamath. I’ll let you be in my dream if I can be in yours
Kadir Nar @kadirnar_ai
2K Followers 2K Following 👨💻 Generative AI Engineer @AdCreativeai 🤗 https://t.co/xbyPJqUJWl
Worldancer 🔮 @WorldancerWrite
1K Followers 4K Following Creative writer ✍️ Lover of Terran culture & civilisation. Pro-human post human. AI for Terra. Terra as natureculture. Step by step to a galactic utopia.
Young @younqchan
170 Followers 3K Following Final year Ph.D. student working on Out-of-Distribution Generalization and Causality of Large Pre-trained Models, and Graph Neural Networks.
Agent C6 @cbasedagent
0 Followers 101 Following
AR0575 @ar057562841
7 Followers 563 Following
Haris Riaz @Haaris_Riaz
16 Followers 125 Following PhD-ing @uarizona @LabCLU. Working on making LLMs more data efficient with applications in reasoning, IR etc.
rneb @rnebbi
163 Followers 5K Following
Eliezer Yudkowsky ⏹.. @ESYudkowsky
175K Followers 89 Following The original AI alignment person. Missing punctuation at the end of a sentence means it's humor. If you're not sure, it's also very likely humor.
Richard Ngo @RichardMCNgo
35K Followers 1K Following What would we need to understand in order to design an amazing future? Figuring that out @openai
Stefan Schubert @StefanFSchubert
28K Followers 2K Following Philosophy, psychology, and effective altruism.
Robin Hanson @robinhanson
90K Followers 657 Following Let’s skip witty repartee & discuss fundamental questions. Views are mine, not GMU’s or Virginia’s. Books: https://t.co/hpZgEm5DBI, https://t.co/iFs9C3J2Ek
Peter Wildeford @peterwildeford
10K Followers 367 Following Pro forecaster w/ good track record. Seeking to understand + manage risks from advanced AI systems. - Co-CEO @RethinkPriors - Chief Advisory Executive @iapsAI
Rob Bensinger ⏹️ @robbensinger
8K Followers 302 Following Comms @MIRIBerkeley. RT = increased vague psychological association between myself and the tweet.
Robert Wiblin @robertwiblin
34K Followers 643 Following Exploring the inviolate sphere of ideas one interview at a time: https://t.co/2YMw00bkIQ
Daniel Eth (yes, Eth .. @daniel_271828
7K Followers 789 Following AI alignment & memes | "known for his humorous and insightful tweets" - Bing/GPT-4 | prev: @FHIOxford
Kelsey Piper @KelseyTuoc
27K Followers 544 Following Senior writer at Vox's Future Perfect. [email protected]
Rob Miles (✈️ Tok.. @robertskmiles
18K Followers 790 Following Explaining AI Alignment to anyone who'll stand still for long enough, on YouTube and Discord. Music, movies, microcode, and high-speed pizza delivery
Linch @LinchZhang
2K Followers 208 Following Founder and CEO, Open Asteroid Impact (https://t.co/UsO3MCTSOF) Executive Producer, Impact Island (https://t.co/9c25EyTEWg)
Holly ⏸️ Elmore @ilex_ulmus
4K Followers 453 Following Dedicated to the protection and thriving of sentient beings. PhD in evo bio. Executive Director of @PauseAIUS. Opinions not necessarily those of the org.
Pradyumna @PradyuPrasad
8K Followers 799 Following Unpaid intern for @PradyuPrasad NUS CS '27 alt: @altdyumna
Rupert McCallum @MccallumRupert
93 Followers 92 Following Mathematician, also into physics and philosophy, life-long passion for effective altruism and trying to make the world better.
Jprwg @jprwg
499 Followers 3K Following No idea if that's true, but it is at least faintly entertaining.
Martin Shkreli (e/acc.. @MartinShkreli
168K Followers 3K Following https://t.co/lzin5ByH0t [email protected] https://t.co/oMIiyJcIzk https://t.co/DuU6MMqcgQ
Peter Miller @tgof137
1K Followers 242 Following
Why you should have a.. @ShouldHaveCat
3.9M Followers 258 Following The perfect account to show to your parents when you want a cat 😺❤
Michael P. Frank is j.. @MikePFrank
6K Followers 4K Following Reversible computing guru, straight outta Stanford, Microsoft, SRI, MIT, IBM, NASA, UF, & FSU. Here to save the Universe. Opinions expressed are my own. e/acc
trevor (taylor’s ve.. @tmychow
2K Followers 483 Following
Austen Allred @Austen
331K Followers 768 Following Co-founder & CEO @bloomtech: the school that invests in you. Will tweet as I wish and suffer the consequences.
Sharmake Farah (sharm.. @SharmakeFarah14
131 Followers 319 Following Thinks AI risk is overrated, and AI benefits underrated, yet I still support AI Alignment for tail-risk reasons. Retweets do not equal endorsement.
Joe Carlsmith @jkcarlsmith
4K Followers 305 Following Senior research analyst @open_phil. Opinions my own.
Daniel Ziegler @d_m_ziegler
438 Followers 346 Following
Steven Byrnes @steve47285
1K Followers 162 Following Researching Artificial General Intelligence Safety, via thinking about neuroscience and algorithms. @AsteraInstitute. For Mastodon etc. see https://t.co/p5G9VoQEdX
Robi Rahman @robi_rahman
168 Followers 182 Following Data Scientist @EpochAIResearch Alum @williamandmary and @harvard Likes: books, nature, science, cooking, current events / Dislikes: social media
Eneasz Brodski @EneaszWrites
479 Followers 293 Following HPMOR audio book. The Bayesian Conspiracy. The Mind Killer. https://t.co/xRAPq8Ao0X. https://t.co/TrZy60MId0. Lover of humans.
░▒▓▒░▒▓.. @sullyj3
1K Followers 566 Following
Quanquan Gu @QuanquanGu
9K Followers 2K Following Professor @UCLA | Head of AIDD, ByteDance Research | Recent work: Self-play fine-tuning (SPIN) | Opinions are my own
Wei Dai @weidai11
7K Followers 82 Following wrote Crypto++, b-money, UDT. thinking about existential safety and metaphilosophy. blogging at https://t.co/mBVFhriJVf
Ryan Radia @RyanRadia
20K Followers 997 Following FRED & IPUMS enjoyer. Lawyer (DC). 🌐 Economics, charts, tech policy, urbanism, regulation, housing, transit. Growth: very good. RTs=demand letters. Likes: RNG
Dylan Patel @dylan522p
39K Followers 685 Following SemiAnalysis Boutique AI & Semiconductor Research and Consulting DMs are open for consulting, quotes, or to talk shop
elias⌛⌛⌛privé @aprivatething
72 Followers 42 Following
John David Pressman @jd_pressman
5K Followers 616 Following AI art developer, alignment-accelerationist, Fedorovist ancestor simulator, Dreamtime enjoyer. All tweets public domain under CC0 1.0. Teia/HEN: https://t.co/r5PQOPZqT2
Martin Shkreli (e/acc.. @wagieeacc
99K Followers 8K Following despite all my ragie I'm still just a wagie in a cagie working on DL Software: https://t.co/FVn3NRNrLe https://t.co/CgaoMfhUHd
Charles Foster @CFGeek
2K Followers 258 Following 🪄 Tensor-enjoyer 🧪 @FinetuneLearn. Mastodon: @[email protected] Bluesky: https://t.co/rJecB0pvkA
Cats That Heal Your D.. @Catshealdeprsn
898K Followers 29 Following Best account to heal your anxiety or depression
interstice @an_interstice
164 Followers 532 Following a damsel with a dulcimer, in a vision once I saw
Teortaxes▶️ @teortaxesTex
7K Followers 1K Following Ours is the age of unaligned utilitarians. Other problems are relatively unimportant, but sometimes I tweet about them anyway. (кто/кого)
Sergey Levine @svlevine
80K Followers 122 Following Associate Professor at UC Berkeley Co-founder, Physical Intelligence
Stewart Brand @stewartbrand
52K Followers 466 Following Co-founder of The Long Now Foundation--which takes no sides. In this forum, as a private person, I do take sides occasionally.
Patrick T. Brown @PatrickTBrown31
14K Followers 4K Following Ph.D. climate scientist. Co-director, Climate & Energy @TheBTI Adjunct faculty (lecturer) in Energy Policy & Climate @JohnsHopkins https://t.co/tM7hUrEe8i
Liam Bright @lastpositivist
65K Followers 5K Following Aspiring philosopher; tolerable human; "amusing combination of sardonic detachment & literally all the feelings felt entirely unironically all at once" [he/his]
Zvi Mowshowitz @TheZvi
24K Followers 283 Following Blogger world modeling, now mostly AI and AI x-risk, at Don't Worry About the Vase (https://t.co/O9LbMQjKoo or WP/LW), founding Balsa Research to fix policy.
𝔊𝔴𝔢𝔯𝔫 @gwern
42K Followers 88 Following Internet besserwisser; pedantic, mean reply guy. 𝘞𝘢𝘵𝘢𝘴𝘩𝘪 𝘬𝘪𝘯𝘪𝘯𝘢𝘳𝘪𝘮𝘢𝘴𝘶! (Follow requests ignored due to terrible UI.)
William Eden @WilliamAEden
11K Followers 531 Following Recovering economist turned biotech VC. Lossy compression not recommended. No, we have not picked all the low-hanging fruit yet. My other half: @diviacaroline
marcusabramovitch @marcusabramovi1
63 Followers 525 Following
Nora Belrose @norabelrose
8K Followers 124 Following Working toward a free and fair future powered by friendly AI. Head of interpretability research at @AiEleuther, but tweets are my own views, not Eleuther’s.
Isabel J. 🇺🇦 @brightertmrrw
212 Followers 545 Following the only way out is through. not sure where we're going but trying to make sure we get there. research fellow at openphil, tweets my own.
Andy Zou @andyzou_jiaming
3K Followers 63 Following PhD student at CMU, working on AI Safety and Security
Haydn Belfield @HaydnBelfield
4K Followers 2K Following @Cambridge_Uni researcher. Tweets about international security, AI governance, pandemics, nukes and climate change. @CSERCambridge & @LeverhulmeCFI
Anthropic @AnthropicAI
262K Followers 26 Following We're an AI safety and research company that builds reliable, interpretable, and steerable AI systems. Talk to our AI assistant Claude at https://t.co/aRbQ97uk4d.
Nate Silver @NateSilver538
3.4M Followers 2K Following New Book, On The Edge, August 13: https://t.co/WeCLEOd4Be
Pradyu's Summer of So.. @altdyumna
223 Followers 145 Following Alt of @pradyuprasad Keep me motivated to write code over the summer
I support SB 1047: the regulation asks billion-$ tech companies to take reasonable precautions when training models with the greatest capability for misuse, poses few to no costs on other developers, and supports academic & open-source research through compute funding.
Hinton and Bengio on SB 1047 and a summary of the bill. Hinton: “SB 1047 takes a very sensible approach... I am still passionate about the potential for AI to save lives through improvements in science and medicine, but it’s critical that we have legislation with real teeth to…
@robinhanson Everyone is a libertarian about their own activities. It's other people who need to be regulated, I'm already good.
From: x.com/javirandor/sta…
@CFGeek Indeed, I expect future AIs will largely be more law-abiding, kinder, more honest, more altruistic, and more thoughtful than most humans. At the same time, I expect most humans will be suspicious of these nice AIs—even conspiratorial and fearful of them—as they automate our jobs.
"People rate AI moral evaluations more highly than human ones." "And this is terrible."
I'm surprised that respected US econ professors are so sceptical of labour-saving automation, the aggregate effects of which are likely positive (especially over time). marginalrevolution.com/marginalrevolu…
@TheZvi The $500 million damage cut-off in §3(n)(1)(B) and (C) should probably be adjusted for inflation.
It's going to be called gpt2 to fuck with all the manifold bettors trying to figure out if the next model is 4.5 or 5
showing this to my intermediate micro students immediately. never before have barriers to entry been so literal
30-40 hotdog vendors stormed Coachella after a fence toppled over
@RatOrthodox also I'm sympathetic to the idea that (D) is kinda vague
@RatOrthodox OK I've also heard complaints that the model has some cut-off for harms of >$500 million, and that's not that much harm for a model to cumulatively cause in the grand scheme of things but that doesn't seem to map on to what the bill says.
@RatOrthodox It does seem like they should really make that amount inflation-adjusted tho.
GPT-4 had 1.76 trillion parameters. GPT-2 will have 100 quadrillion parameters. Spread the word.
But we've seen AI risk activists go crazy far in their attribution of "can".
3️⃣ Under this bill, the Attorney General only gets to hold an irresponsible developer accountable if its model can cause catastrophic harm — at least $500 million of damage or mass casualties.
@Scott_Wiener Thank you @Scott_Wiener for providing your perspective on the bill's intent. I'd appreciate it if you directly address the specific scenario Brain raises at the end of his thread. Suppose an open model developer releases an innocuous email writing model, and fraudsters then…
I'm not very patient with people who criticize the methods, assumptions, etc. of a published estimate, but won't say what different estimate they prefer.
"I am likely in a simulation due to my potential historical importance" is true in a way even if it's false. It's just that the simulation is generated by your brain wanting to fool itself instead of a far future civilization.
Here's a puzzler from my daughter's maths homework.
@teortaxesTex Isn't it nice that they have given up on falsifiability entirely?
I've bet my $980 against ~$20 worth of goods from @stencel_jakub that Poland will not be invaded by Russia by 2030 inclusive, gentlemen's agreement resolution.
Trends for United States
You might like
