Florian Mai 🇺🇳 @_florianmai
Postdoc at the Machine Learning / Language Intelligence and Information Retrieval group @CW_KULeuven. PhD from @EPFL_en. florianmai.github.io Joined December 2013-
Tweets4K
-
Followers1K
-
Following1K
-
Likes5K
Predictions: >=2 orgs will get 35% on SWE-bench by Aug 1, 2024. A fully open source system will reach 35% by Nov 1, 2024. Probably based on SWE-agent + ACI improvements: debugger, better code retrieval, lang. server protocol. The LM will be finetuned on ~500 good trajectories
Introducing the Instruction Hierarchy, our latest safety research to advance robustness for prompt injections and other ways of tricking LLMs into executing unsafe actions. More details: arxiv.org/abs/2404.13208
🚨 @geoffreyhinton warning us about AI x-risk with growing urgency Specifically names OpenAI and Meta Low-integrity @pmarca tries to name-call everyone as "baptist or bootlegger" 🙄 Hinton breaks his asinine dichotomy; retired godfather of AI has nothing left to prove or earn.
"Planning is a less mature technology, and I find it hard to predict in advance what it will do" this sounds like catastrophic events/accidents with advanced AI are quite possible, contrary to what he's been claiming🤨
"Planning is a less mature technology, and I find it hard to predict in advance what it will do" this sounds like catastrophic events/accidents with advanced AI are quite possible, contrary to what he's been claiming🤨
Microsoft presents Rho-1 Not All Tokens Are What You Need Previous language model pre-training methods have uniformly applied a next-token prediction loss to all training tokens. Challenging this norm, we posit that "Not all tokens in a corpus are equally important for
Both Claude and Gemini 1.5 aren't available in the EU. Is this due to regulatory issues? If so and this problem persists, the EU will soon lag behind the US in productivity. It may not be a problem long-term, but in the medium term this will affect the economy and thus politics.
I'm skeptical that Chatbot Arena is really as informative as people make it out to be, but I'd be glad to learn that I am wrong: 1. Different chatbots have really distinct talking styles. Isn't it easy to tell whether something comes from GPT-4 or Grok? Then it's not really…
I'm skeptical that Chatbot Arena is really as informative as people make it out to be, but I'd be glad to learn that I am wrong: 1. Different chatbots have really distinct talking styles. Isn't it easy to tell whether something comes from GPT-4 or Grok? Then it's not really…
Why are Transformers considered better than RNNs + self-attention + FFN? Sure, Transformers can parallelize training across the sequence dimension. But with an RNN you can still parallelize across the batch dimension to maximize utilization of your GPU. Why is that not enough?
Bad news. France opted to host an AI Safety Summit in November 2024, but several of our sources confirm it has been postponed to February 2025. It has also been renamed to “AI Action Summit”, dropping the all-important safety focus. Safety will be a minor part of the summit,…
In light of MS+OAI investing $100B into a new data center, $2B for all of Canada seems insignificant, let alone only $50M for AI Safety. AGI is likely the last human invention ever, and many companies are out to build it. If you want to have a say, you have to invest much more.
In light of MS+OAI investing $100B into a new data center, $2B for all of Canada seems insignificant, let alone only $50M for AI Safety. AGI is likely the last human invention ever, and many companies are out to build it. If you want to have a say, you have to invest much more.
@hereisramji @deedydas The real scandal here is that this post has 13.5k likes and your clarification post has 20. Media literacy is fundamentally broken. No wonder people around the world increasingly vote demagogues into power.
“A world without nuclear weapons is totally possible,” says @Emma_Pike_, a nuclear disarmament consultant and activist. She shared her thoughts with Times Opinion on TikTok: nyti.ms/49buYeZ
@SmokeAwayyy This narrative is unsupported. Multiple people involved in the matter have stated that it was not about AI safety, including the interim CEO @eshear , who was appointed by the board that fired Altman at the time. Don't spread conspiracy theories. This is only damaging.
A Russia-US #NuclearWar would be humanity's dumbest act yet & may kill about 99% in the US, Russia and Europe as seen in our simulation:
So excited and so very humbled to be stepping in to head AI Safety and Alignment at @GoogleDeepMind. Lots of work ahead, both for present-day issues and for extreme risks in anticipation of capabilities advancing.
So excited and so very humbled to be stepping in to head AI Safety and Alignment at @GoogleDeepMind. Lots of work ahead, both for present-day issues and for extreme risks in anticipation of capabilities advancing.
It's a great relief that no one solved language modeling before the COLM abstract deadline. I was getting kind of worried in the fall.
Introducing Evolutionary Model Merge: A new approach bringing us closer to automating foundation model development. We use evolution to find great ways of combining open-source models, building new powerful foundation models with user-specified abilities! sakana.ai/evolutionary-m…
Excited to share something that we've needed since the early open RLHF days: RewardBench, the first benchmark for reward models. 1. We evaluated 30+ of the currently available RMs (w/ DPO too). 2. We created new datasets covering chat, safety, code, math, etc. We learned a lot.…
The destructive ambition of many, that started with the use of a rock to crack nuts instead of using your teeth, that caused the mass unemployment of nut-crackers in 14'500BC.
The destructive ambition of many, that started with the use of a rock to crack nuts instead of using your teeth, that caused the mass unemployment of nut-crackers in 14'500BC.
AK @_akhaliq
310K Followers 3K Following AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo follow on Hugging Face: https://t.co/q2Qoey80Gx
Kyunghyun Cho @kchonyc
61K Followers 2K Following a combination of a mediocre scientist, a mediocre manager, a mediocre advisor & a mediocre PC at @nyuniversity (@CILVRatNYU) & @genentech (@PrescientDesign).
François Fleuret @francoisfleuret
31K Followers 456 Following Prof. @Unige_en, Adjunct Prof. @EPFL_en, Research Fellow @idiap_ch, co-founder @nc_shape. AI and machine learning since 1994. I like reality.
Felix Hill @FelixHill84
9K Followers 777 Following Research Scientist, Deepmind I try to think hard about everything I tweet, esp on 90s football and 80s music None of my opinions are really someone else's
Sam Bowman @sleepinyourhat
35K Followers 3K Following AI alignment + LLMs at NYU & Anthropic. Views not employers'. No relation to @s8mb. I think you should join @givingwhatwecan.
Pasquale Minervini �.. @PMinervini
7K Followers 4K Following Researcher in ML/NLP at the University of Edinburgh (faculty @InfAtEd @EdinburghNLP), @ELLISforEurope, @UCL_NLP, PI for @Clarify2020, https://t.co/WydvfU8ugz he/they
Sebastian Ruder @seb_ruder
80K Followers 1K Following Multilingual LLMs @cohere • Prev: @GoogleDeepMind • Newsletter: https://t.co/7JGh2qpG98
Leo Boytsov @srchvrs
7K Followers 2K Following Sr. Research Scientist @AWS Labs (ph-D @LTIatCMU) working on unnatural language processing, speaking πtorch & C++. Opinions sampled from MY OWN 100T param LM.
Angelos Katharopoulos @angeloskath
2K Followers 236 Following Machine Learning Research @Apple. Previously PhD student at @idiap_ch and @EPFL. Interested in all things machine learnable
Jim Fan @DrJimFan
229K Followers 3K Following @NVIDIA Sr. Research Manager & Lead of Embodied AI (GEAR Lab). Creating foundation models for Humanoid Robots & Gaming. @Stanford Ph.D. @OpenAI's first intern.
Ofir Press @OfirPress
10K Followers 3K Following I build tough benchmarks for LMs and then I get the LMs to solve them. Postdoc @Princeton. PhD from @nlpnoah @UW. Ex-visiting researcher @MetaAI & @MosaicML.
Sebastian Riedel (@ri.. @riedelcastro
15K Followers 470 Following Researcher in NLP/ML @deepmind, @ucl_nlp, @[email protected] on Mastodon
rohan anil @_arohan_
12K Followers 2K Following Principal Engineer, @GoogleDeepMind Gemini. prev PaLM-2. Tinkering with optimization and distributed systems. opinions are my own.
Leshem Choshen 🤖�.. @LChoshen
4K Followers 550 Following 🥇 Collaborative LLMs 🥈 Opinionatedly sharing #ML & #NLP 🥉 Propagating us underdogs we owe science an alternative hype @IBMResearch & @MIT_CSAIL
Suraj Srinivas @Suuraj
926 Followers 984 Following Postdoc @harvard / PhD @epfl_en / Bangalorean 🇮🇳 / trying to understand deep learning
Ashutosh Mehra @ashutoshmehra
1K Followers 5K Following Senior Principal Scientist at Adobe. Working on Acrobat AI Assistant, LLMs, and document ML.
Kondwani Ngulube @Kondwani_G
24 Followers 91 Following software engineer 💻 @alueducation 22 building @shambaData
George Lloyd-King @GeorgeLloydKing
700 Followers 2K Following Communications Manager @AdaLovelaceInst
For Humanity Podcast .. @ForHumanityPod
737 Followers 2K Following The accessible AI safety podcast for all, no tech background necessary. Focused only on human extinction risk #alignment #interpretability #ai #aisafety
Roman Kalkreuth 🏳�.. @RomanKalkreuth
253 Followers 198 Following Assistant Professor (Akademischer Rat) at the Chair for AI Methodology of RWTH Aachen University (Germany)
Ajay Jain @ajayj_
6K Followers 3K Following Co-founder @genmoai. Co-created denoising diffusion (DDPM), DreamFusion, Dream Fields. Ex Ph.D. @berkeley_ai, @googleai, @facebookai, @nvidiaai, @mit
Toby Lightheart @TobyLightheart
147 Followers 969 Following Learning machines. Interested in AGI, philosophy and education. PhD in Engineering (neural networks).
Chryssa Zerva @chryssaZrv
378 Followers 325 Following Assistant Professor @informatica_IST, @ist_tecnico. Interested in understanding uncertainty in data, models, life. NLP, ML and climbing fan.
filippo @filippo_v
328 Followers 2K Following
Phily8020 @phily8020
108 Followers 936 Following ACCELERATE AI! AI researcher experimenting with LLM's. Chronically Asking Questions.
Nino Scherrer @ninoscherrer
585 Followers 2K Following Interested in rigorous LLM evals, synthetic data, causality, robustness & AI safety/ethics | RS @PatronusAI | Prev: @VectorInst, @Mila_Quebec, @MPI_IS, @ETH_en
Profoundlyyyy @profoundlyyyy
4K Followers 4K Following We should be thoughtful about this AI thing. Hope to share boldly, be wrong sometimes, and learn
Tara Steele ⏸️ @tarasteele22
52 Followers 130 Following ‘…there’s 10% chance AI will wipe out humanity in the next 20 years’, Geoffrey Hinton, ‘Godfather of AI’. Perhaps stopping for a bit would be sensible?!🤷🏼♀️
corto @corto02982478
239 Followers 3K Following
Burny — Effective O.. @burny_tech
14K Followers 6K Following Transhuman engineer in singularity! Lover of AI & omnidisciplionary metamathemagics! Hypercuriousia! Omniperspectivity! Shapeshifting metafluid! Freedom 4 all!
Alpaca 🦙 (e/delve) @LeetAlpaca3
211 Followers 3K Following Servant of the future. If I blocked you, I probably didn’t mean to. (Big block list from long ago) Tweets are my own.
Dr. Peter S. Park ⏸.. @dr_park_phd
1K Followers 781 Following AI Existential Safety Postdoctoral Fellow @MIT, @Tegmark Lab. @Harvard PhD '23, @Princeton '17. Alum of @JoHenrich Lab. Studies cognition (both human and AI).
Information MDPI @InformationMDPI
1K Followers 2K Following Information (ISSN 2078-2489, #Scopus, #ESCI, #EI Compendex) is an open access journal of information science and technology, data, knowledge and communication.
Nash Pat @NashPat2
63 Followers 390 Following
Dimitris Proios @ProiosDimitris
8 Followers 248 Following
carlo @carlo_l_fritz
300 Followers 4K Following econ/sociology undergrad, unintentionally (edit: steadily) falling into the methodological stuff... wish me luck! || turtles all the way down
Tr @trx_3333
42 Followers 539 Following
Billy Vythikowski @vythikowski
33 Followers 314 Following
aosuenth @aosuenth
105 Followers 45 Following
Muhammad Suleman Asif @msulemanas57411
252 Followers 6K Following Current :-Senior Analytic Consultant @wellsfargo. Previously :-Founder of WIFC (Without Internet free Call). I go by Muhammad.
Øystein Runde @OysteinRunde
812 Followers 2K Following A podcast about possible futures (Wunderdog) 13 books (comics!) out in Norway, often weird and somewhat research-heavy Writer/Artist https://t.co/d7FprSC0i0
Ruairi @ruairiSpain
263 Followers 2K Following
M. ElNokrashy @__munael
134 Followers 3K Following Applying science in AI and Informatics 🤖. Reading stories, writing some. Personal account; employer not involved; RT =/= Endorsement; etc.
Markus Rauhalahti, Ph.. @MRauhalahti
844 Followers 5K Following Independent researcher: molecular design/nanotech, human-AI collab, informetrics, DIY-scihw. Prev compchem/biophys/info phd&postdoc @helsinkiuni
Torsten Jacobi @jacobi_torsten
1K Followers 2K Following Serial Entrepreneur. Building intelligent AI Agents that automate the economy. See if we cover your industry workflow already https://t.co/vgJs0wywEu.
Atif_735 @tiger_1724
18 Followers 436 Following
Segmond Yunsai @ysegmond
335 Followers 604 Following Interests:- wrenching old bmws vROOOM, programming vRAAAM
Michael Huang ⏸️ @michhuan
345 Followers 252 Following Will we eradicate involuntary suffering and start the @postsuffering era? (And will we @PauseAI unless its safety is proven?)
Ross Greer @Ross__Greer
38 Followers 320 Following AI & CV Research at UCSD ~ intelligent vehicles with https://t.co/N4FmD93H8B ~ ~ machine learning for music & audio with https://t.co/A8bTVmLos5 ~
Hasnain Bukhari @OhHasnain
468 Followers 854 Following A little bit of slope makes up for a lot of y-intercept. Formerly @wise, @revolutapp wealth & trading. Building @go_lightyear.
Mirco Ravanelli @mirco_ravanelli
4K Followers 2K Following Deep learning for Conversational AI. Creator of SpeechBrain.
axel kramer (@axkra@m.. @axkra
136 Followers 1K Following working on a sketching app to help me think. love designing and writing software. broad background: research & tech & finance companies. @[email protected]
Paul C. Jeffries @PaulJeffries
921 Followers 2K Following FounderPool & Venture Capital & GBBK board (now), Facebook (back then), Physics & Philosophy (way back then), Ideas are Toys (always) https://t.co/4mdMUtcw3w
Silicon Heart @heart_silicon
1 Followers 44 Following Weekly dispatch on tech, humanity and occasional weird stuff
JDKee @jdkee
3 Followers 2K Following dat://399632a7162d0bd779538725da0edc4f89d98483063a808ad75cf58926e9b41f/
Oya Aran @aranoya
312 Followers 2K Following PhD Computer Eng. @UniBogazici, Social Computing, Machine Learning, Comp. Vision Researcher, Mom, Sailor, Potter
(((ل()(ل() 'yoav))).. @yoavgo
46K Followers 2K Following
Yann LeCun @ylecun
710K Followers 718 Following Professor at NYU. Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.
Andrej Karpathy @karpathy
978K Followers 904 Following 🧑🍳. Previously Director of AI @ Tesla, founding team @ OpenAI, CS231n/PhD @ Stanford. I like to train large deep neural nets 🧠🤖💥
AK @_akhaliq
310K Followers 3K Following AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo follow on Hugging Face: https://t.co/q2Qoey80Gx
elvis @omarsar0
189K Followers 486 Following Building with LLMs @dair_ai • Prev: Meta AI, Galactica LLM, PapersWithCode, Elastic, PhD • Creator of the Prompting Guide (~4M learners)
Sebastian Raschka @rasbt
267K Followers 906 Following Machine learning & AI researcher writing at https://t.co/A0tXWzG1p5. LLM research engineer @LightningAI. Previously stats professor at UW-Madison.
François Chollet @fchollet
469K Followers 770 Following Deep learning @google. Creator of Keras. Author of 'Deep Learning with Python'. Opinions are my own.
Lucas Beyer (bl16) @giffmana
56K Followers 444 Following Researcher (Google DeepMind/Brain in Zürich, ex-RWTH Aachen), Gamer, Hacker, Belgian. Mostly gave up trying mastodon as [email protected]
Kyunghyun Cho @kchonyc
61K Followers 2K Following a combination of a mediocre scientist, a mediocre manager, a mediocre advisor & a mediocre PC at @nyuniversity (@CILVRatNYU) & @genentech (@PrescientDesign).
François Fleuret @francoisfleuret
31K Followers 456 Following Prof. @Unige_en, Adjunct Prof. @EPFL_en, Research Fellow @idiap_ch, co-founder @nc_shape. AI and machine learning since 1994. I like reality.
Felix Hill @FelixHill84
9K Followers 777 Following Research Scientist, Deepmind I try to think hard about everything I tweet, esp on 90s football and 80s music None of my opinions are really someone else's
AI at Meta @AIatMeta
531K Followers 255 Following Together with the AI community, we are pushing the boundaries of what’s possible through open science to create a more connected world.
Sam Bowman @sleepinyourhat
35K Followers 3K Following AI alignment + LLMs at NYU & Anthropic. Views not employers'. No relation to @s8mb. I think you should join @givingwhatwecan.
Christopher Manning @chrmanning
126K Followers 115 Following Director, @StanfordAILab. Assoc. Director, @StanfordHAI. Founder, @stanfordnlp. Prof. CS & Linguistics, @Stanford. IP @aixventureshq. 🇦🇺 Do #NLProc & #AI. 👋
Tal Linzen @tallinzen
16K Followers 894 Following Professor @nyuling and @NYUDataScience, research scientist @GoogleAI
Richard Socher @RichardSocher
101K Followers 970 Following CEO @youSearchEngine Investing at @aixventuresHQ Before: Stanford Adj Prof in AI/NLP, Chief Scientist at Salesforce, MetaMind
Pasquale Minervini �.. @PMinervini
7K Followers 4K Following Researcher in ML/NLP at the University of Edinburgh (faculty @InfAtEd @EdinburghNLP), @ELLISforEurope, @UCL_NLP, PI for @Clarify2020, https://t.co/WydvfU8ugz he/they
Yi Tay @YiTayML
29K Followers 97 Following chief scientist / cofounder @RekaAILabs 🫠 past: research scientist @google brain 🤯 currently learning to be a dad 🍼
Hannah Rose Kirk @hannahrosekirk
3K Followers 683 Following AI researcher trying to make sense of all things cyberspace 🤖 Uni of Ox PhD (loading…) @oiioxford & @OxfordAI. Prev @turinginst & @Cambridge_Uni. Visitor @ NYU
Brian Roemmele @BrianRoemmele
269K Followers 31K Following we can only see what we think is possible...
Allan Dafoe @AllanDafoe
3K Followers 562 Following AGI governance: navigating the transition to beneficial AGI (Google DeepMind)
For Humanity Podcast .. @ForHumanityPod
737 Followers 2K Following The accessible AI safety podcast for all, no tech background necessary. Focused only on human extinction risk #alignment #interpretability #ai #aisafety
Roman Kalkreuth 🏳�.. @RomanKalkreuth
253 Followers 198 Following Assistant Professor (Akademischer Rat) at the Chair for AI Methodology of RWTH Aachen University (Germany)
Chryssa Zerva @chryssaZrv
378 Followers 325 Following Assistant Professor @informatica_IST, @ist_tecnico. Interested in understanding uncertainty in data, models, life. NLP, ML and climbing fan.
davidad 🎇 @davidad
13K Followers 7K Following Programme Director @ARIA_research | accelerate mathematical modelling with AI and categorical systems theory » build safe transformative AI » cancel heat death
Andrew Curran @AndrewCurran_
11K Followers 7K Following Atypically Friendly - I write about AI and human creativity. Will periodically make extremely unusual arguments.
udio @udiomusic
28K Followers 0 Following
Aidan Clark @_aidan_clark_
4K Followers 210 Following Research @OpenAI. Ex: @DeepMind, @BerkeleyDAGRS Hae sententiae verbaque mihi soli sunt
Ashwinee Panda @PandaAshwinee
944 Followers 602 Following PhD @princeton, @Cal alum, currently working on LLMs
James Bradbury @jekbradbury
11K Followers 8K Following Compute at @AnthropicAI! Previously JAX, TPUs, and LLMs at Google, MetaMind/@SFResearch, @Stanford Linguistics, @Caixin.
Tara Steele ⏸️ @tarasteele22
52 Followers 130 Following ‘…there’s 10% chance AI will wipe out humanity in the next 20 years’, Geoffrey Hinton, ‘Godfather of AI’. Perhaps stopping for a bit would be sensible?!🤷🏼♀️
Dr. Peter S. Park ⏸.. @dr_park_phd
1K Followers 781 Following AI Existential Safety Postdoctoral Fellow @MIT, @Tegmark Lab. @Harvard PhD '23, @Princeton '17. Alum of @JoHenrich Lab. Studies cognition (both human and AI).
Ivan Habernal @ivanhabernal
629 Followers 61 Following Full professor at @ruhrunibochum + @RCTrustworthy | Leading the TrustHLT group | Ex @TUDarmstadt @unipb | Playing the bass | He/him/his
Risto Uuk @RistoUuk
2K Followers 1K Following EU Research Lead at Future of Life Institute (@FLI_org) focused on researching European policy-making on AI
near @nearcyan
45K Followers 883 Following https://t.co/IdaJwZJCXm partner @ https://t.co/9g1MIgjiqc dms open
SynthLabs @synth_labs
12K Followers 43 Following AI Aligned with Your Vision. We’re doing cutting edge research for transparent, auditable AI alignment.
Frontier Model Forum @fmf_org
122 Followers 5 Following A coalition of industry leaders, committed to the safe and responsible development of frontier AI
Stefan Baumann @StefanABaumann
108 Followers 135 Following PhD Student @ Ommer Lab/CompVis (@LMU_Muenchen, @ELLISforEurope) working on generative computer vision
Nathan Lambert @natolambert
25K Followers 689 Following Figuring out AI @allen_ai, "rl boi" DM me papers. Writes @interconnectsai, talks @retortai Has phd and some credentials
Andreas Waldis @AndreasWaldis
66 Followers 78 Following PhD Student at UKP Lab, TU Darmstadt/Hochschule Luzern
Mirco Ravanelli @mirco_ravanelli
4K Followers 2K Following Deep learning for Conversational AI. Creator of SpeechBrain.
Markus Anderljung @Manderljung
2K Followers 768 Following Trying to design good AI policy. Head of AI Policy & Research Fellow @GovAI_, Adjunct Fellow @CNASdc
Darren McKee @dbcmckee
649 Followers 568 Following Author of newly released "Uncontrollable" - a beginner friendly AI safety book https://t.co/Ib0CKIK2pq Advisor | Speaker | Host of The Reality Check
Tomáš Daniš @tmdanis
617 Followers 358 Following Natural intelligence researching artificial intelligence. Tweets about AI, alignment, tech, language learning and anything else I find interesting
Aaron Bergman 🔍 �.. @AaronBergman18
2K Followers 1K Following 👎: suffering | 👍: EA, AI alignment, decoupling, R, cringe, amateur pharmacology + programming | Georgetown ‘22 (math+econ+phil) | Career status: 🤷♂️
Adele Goldberg @adelegoldberg1
7K Followers 2K Following Linguist, experimentalist, constructionist usage-based approach to language (she's)
Leonie Weissweiler @LAWeissweiler
788 Followers 314 Following Visiting Researcher with @adelegoldberg1 at @Princeton | prev. @cislmu @LTIatCMU @CambridgeLTL
Tonglet Jonathan @TongletJ
94 Followers 547 Following 🇧🇪 ELLIS PhD student @ TU Darmstadt and KU Leuven. Multimodal learning and Fact-checking.
Martin Fajčík @martin_fajcik
40 Followers 99 Following NLP Researcher at BUT-FIT https://t.co/gcLnaWwco3
Tetraspace 💎🔎 @TetraspaceWest
5K Followers 2K Following here to believe true things and do good actions 💎 let's solve AI alignment 💎 enjoying things rules 💎 banner from lesswrong
Brian Atwood @batwood011
4K Followers 2K Following Creating conversational speech AI for the real world @sindarintech. Try for yourself at https://t.co/3HgjKAswWz and https://t.co/Pu2Rmho3JI
Karol Hausman @hausman_k
22K Followers 141 Following @Physical_int ex: researcher @GoogleAI/@DeepMind, adj. Prof. @Stanford. Into robots, AI, NBA, philosophy, soccer and almond croissants. 🇵🇱🇺🇸
Molly ⏸️ Hickman @celloMolly
388 Followers 595 Following Data scientist @ Forecasting Research Institute. Formerly @ nLine, Inc. measuring power quality at outlets around the world. 🧬 Samotsvety Forecasting
Haydn Belfield @HaydnBelfield
4K Followers 2K Following @Cambridge_Uni researcher. Tweets about international security, AI governance, pandemics, nukes and climate change. @CSERCambridge & @LeverhulmeCFI
Ben Dickson @bendee983
4K Followers 599 Following Software Engineer | Tech analyst | Thinker | Student of life | Founder of @bdtechtalks
Jeremie Harris @jeremiecharris
4K Followers 570 Following Co-founder & CEO of Gladstone AI We promote responsible AI R&D and adoption by designing and deploying safeguards against AI-driven national security threats.
Edouard Harris @harris_edouard
5K Followers 2K Following Cofounder & CTO @GladstoneAI https://t.co/aGM324ATKG
Kabir Kumar, AI-plans.. @KKumar_ai_plans
264 Followers 325 Following Views are my own. Critique AI Alignment plans at https://t.co/zxHhGCbtFI I'm stupid, impatient and unemployed
Mario Cannistrà @Blueyatagarasu
175 Followers 68 Following This is the most important time in history for humanity to be wise. We must solve the AI alignment problem.
Shaun Ralston @shaunralston
1K Followers 2K Following @OpenAI @SutterHealth @Webvan | BusDev – e/acc. Sonoma County Aficionado. Cyclist. Sticky Bun Seeker. AI Blogger. Technologist. Music Lover. Libertarian.
Grace Kind @kindgracekind
2K Followers 2K Following AI navel-gazer / Ideonomy evangelist / navigator of uncertain waters
Luca Bertuzzi @BertuzLuca
13K Followers 4K Following Technology journalist specialised in digital policy & European affairs. Ex @EURACTIV. Bylines @PrivacyPros, @repubblica, @tagesspiegel. DMs open.
Joep Meindertsma ⏸ @joepmeindertsma
469 Followers 480 Following Loves democracy, open data and technology. Building software at @ontola_io and https://t.co/JmDbk7GX7U. Founded @PauseAIPoliticians need to stop patting themselves on the back prematurely, and work towards a binding treaty. It's far more difficult, but it's the thing we need. Be the adults in the room. Don't fuck this up.
Scoop (now free to view): Rishi Sunak’s AI Safety Institute is failing to test the safety of most leading AI models like GPT-5 before they’re released — despite heralding a “landmark” deal to check them for big security threats 👇 politico.eu/article/rishi-…
who could possibly have foreseen this turn of events
Scoop (now free to view): Rishi Sunak’s AI Safety Institute is failing to test the safety of most leading AI models like GPT-5 before they’re released — despite heralding a “landmark” deal to check them for big security threats 👇 politico.eu/article/rishi-…
Seeing what we were able to do in SWE-agent by 'just' building infra on top of an LM, makes me believe that those who also have the power to train/finetune the LM will be able to build systems that are *much* more impressive.
Anyone who thinks that we are near the peak of the performance curve is wrong. These types of agent systems will be able to do much more than GPT-4 can today.
When GPT-4 came out everyone was surprised at the scale of the quality increase from GPT-3 The GPT-5 release will not be as surprising: In SWE-agent we show that building agent infra on top of an LM increases perf from 0% to 12% on a very tough task GPT-5 will be the agent GPT
Predictions: >=2 orgs will get 35% on SWE-bench by Aug 1, 2024. A fully open source system will reach 35% by Nov 1, 2024. Probably based on SWE-agent + ACI improvements: debugger, better code retrieval, lang. server protocol. The LM will be finetuned on ~500 good trajectories
As LMs become much cheaper, it'll be a no-brainer to start running a process like SWE-agent automatically for every bug report. I can't wait to see how software engineering looks in 1 year! I think exciting things will happen :)
The better the bug report (more examples of code bits failing, a better description of your setup env) the higher the probability that SWE-agent will manage to fix it.
As we get better and better at this task, and as LMs become faster and cheaper, programmers will become much more productive and I think their job will change from mainly solving bugs to maybe mainly filing high-quality bug reports.
35% on SWE-bench is super hard but if we've managed to get so far with models that were never trained on anything similar, I think putting similar data (good SWE-agent trajectories) into training is really going to catapult performance.
Literally this, but as though articulated by a children's television show host.
The most insidious form of climate denial is no longer, "It's not happening," but the belief that incremental or tech solutions will solve this crisis. 1/6
GPT4 is almost two years old. In LLM age, that's probably almost a century. Numerous companies are following suit and building LLMs that are almost at the same level as GPT4. What is the reason that OpenAI is keeping us waiting. Do they have nothing to show? My guess: the exact…
I think people should just go back to using linear models then…
We have finally done it. After all this time and due to countless requests from our users, we've shipped what I think is our most important and revolutionary feature yet. You can now interrupt Claude's yapping with our new stop generation button!
Curiosity-driven Red-teaming for Large Language Models "Recent works automate red teaming by training a separate red team LLM with reinforcement learning (RL) to generate test cases that maximize the chance of eliciting undesirable responses from the target LLM. However, current…
New Anthropic research: we find that probing, a simple interpretability technique, can detect when backdoored "sleeper agent" models are about to behave dangerously, after they pretend to be safe in training. Check out our first alignment blog post here: anthropic.com/research/probe…
Persuaded or manipulated by AI? Check out this new paper from @GoogleDeepMind on definitions and mitigations. It was a privilege to advise on this important research! #AIethics #PersuasionAI
🔮 New Google DeepMind paper exploring what persuasion and manipulation in the context of language models. 👀 Existing safeguard approaches often focus on harmful outcomes of persuasion. This research argues for a deeper examination of the process of AI persuasion itself to…
Why are the best minds in AI walking away from Big Tech?
🧵 Another case-in-point. Phi-3 model with only 3.8B params and 3.3T tokens rivals Mixtral 8x7B and GPT-3.5. "The innovation lies *ENTIRELY* in our dataset for training, a scaled-up version of the one used for phi-2, composed of heavily filtered web data and synthetic data." ↩️
In machine learning there are three key factors that affect your model performance: data, data, data.
Trends for United States
You might like
