tegmark pfp

tegmark

@ejio143612

315 Following
252 Followers
tegmark pfp
tegmark
@ejio143612
Two of the funniest things with this resolution are 1. Germany going around convincing Nato countries to abstain & then panicking last minute about how bad it looks for themselves & therefore flipping to yes last minute, leaving countries like Sweden & Belgium looking so stupid
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
One of my favorite 2024 physics papers: a clever new idea for catching the elusive graviton particles we think make up the gravitational field. https://t.co/xr3wGaoOG2
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
If you cluster language model features (from SAEs) into two groups based on whether they tend to fire together in the same document, you find two lobes of features that also turn out to be geometrically distinct! Math vs. prose features separate in this t-SNE plot... https://t.co/zubb2t7pzR
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
RELEASE: THE COMPENDIUM Several reckless groups are racing towards AGI, and risking the extinction of humanity. What is AGI? Who are these people? Why are they doing this? And what can we do about it? We answer these questions in The Compendium. 1/17 https://t.co/EiuC5u0fK8
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
1/6 Based on our work in DafnyBench, we built a VS Code Extension, which we call Dafny Autopilot.
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
1/6 New paper! “The Geometry of Concepts: Sparse Autoencoder Feature Structure.” We find that the concept universe of SAE features has interesting structure at three levels: 1) “atomic” small-scale, 2) “brain” intermediate-scale, and 3) “galaxy” large-scale! https://t.co/tvJ6Gspf1y
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
As autumn falls, Harvard scores a chilling -21.58 in the 2025 Free Speech Ranking. Speech suppression? No, just “fun-sized” freedoms! https://t.co/2JPG8klDzD
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
RT @DavidSKrueger: AI safety trilemma... you can't simultaneously: 1) Think AGI is coming soon 2) Not want to slow it down 3) Be one of th…
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
RT @AISafetyMemes: NYT: The Shoggoth is the “the most important meme in AI” What’s happening in AI today feels, to some of its participan…
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
Slightly schizophrenic report from Citigroup. AGI is arriving in 2029, with ASI soon after. But don't worry: work on your critical thinking, problem solving, communication and literacy skills for a durable competitive advantage. https://t.co/UiBKz9dhXJ
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
RT @AISafetyMemes: Microsoft CEO is openly admitting recursive self-improvement has begun Hey @ericschmidt, you said this was the red line…
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
1/11: New paper! Decomposing the Dark Matter of Sparse Autoencoders. We find that SAE errors and error norms are linearly predictable using model activations. Why is this, and what does it imply for SAE scaling and the structure of language model representations? Answers in 🧵 https://t.co/inWQ7oX0T7
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
Does this help explain why the US Government has been dragging its feet over banning deceptive deepfakes? The Pentagon Wants to Use AI to Create Deepfake Internet Users https://t.co/etrDZgrfg9
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
Sam Altman: if we can make an AI system that is materially better than all of OpenAI at doing AI research, that does feel like an important discontinuity... the model is going to get so good so fast... plan for the model to get rapidly smarter https://t.co/MD2yCkbGWm
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
Many AI pundits have US-China AI collaboration totally backwards. The point is *not* to convince the other to shun uncontrollable AGI, but to jointly ensure that the rest of the world follows suit, including weak or corrupt governments. ? It's not in the US self-interest to https://t.co/Dp9zALUFyQ
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
RT @TolgaBilge_: If you're looking for an alternative to the suicide race proposed in Aschenbrenner's Situational Awareness or Amodei's r…
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
We need more tool AI as in Elon's space triumph today, and less AGI world domination hype as in Dario Amodei's recent entente essay. My essay below argues that scaling quickly won't lead to Dario's eternal 1991 – but perhaps to 1984 until the end, with a non-human Big https://t.co/A68MMJWAeZ
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
💰 🏫 Our PhD and postdoc fellowship applications are open! 🆕 In addition to our technical AI safety fellowships, we have a new track for US-China AI governance PhD fellows. 📆 Apply by 20 November for PhD fellowships; 6 January for postdocs: https://t.co/qbrJygFV3Y https://t.co/jzpXKq7gsS
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
Congratulations on the Nobel Prize, ?@demishassabis? & John Jumper! AlphaFold is not only a revolutionary breakthrough, but also an inspiring example of the AI wonders we can do without building uncontrollable AGI. 🎉🥳 https://t.co/5sKyk6hqc2
0 reply
0 recast
0 reaction
tegmark pfp
tegmark
@ejio143612
RT @tsarnick: Max Tegmark says crazy things will happen due to AI in the next 2 years so we can no longer plan 10 years into the future and…
0 reply
0 recast
0 reaction