Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 16/10/2026RT @deepfates: Uhh could we turn codebases into neural networks and then do verifiable interpretability on them? 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 16/10/2026RT @tenobrus: i mean that was pretty much his claim 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 15/10/2026RT @ladymcscope: What is this extremely mysterious neuron that set mechinterp going on a dead end for the last 3 years? It's literally </\> kiki ? or bouba o0o 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 15/10/2026RT @ctjlewis: I can’t quote Nick’s post but there is a fascinating level of detail in this thing he has written. I don’t think it should be shortened or any details should be removed from this like some comments say, he should just staple a short version to the front. 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 15/10/2026RT @jfonsecarivera: genuinely an amazing day 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 15/10/2026RT @_chenglou: Handwired bridges the gap between traditional code and a neural net in a funny way. It turns if-else and loops into weights, like so: 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 11/10/2026RT @Willob: Grok please make this a TBPN graphic with the title “MARRIED” (Congrats Cat and Ryan ❤️) 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 09/10/2026RT @michael_nielsen: Vinge's "opaque wall across the future", which he described in his 1993 Singularity essay, is now being felt by more and more people... 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 08/10/2026RT @balesni: Two other safety researchers and I were fired from OpenAI last week. We wrote this letter to leadership. I believe we were fired for prioritizing safety over the near-term interests of OpenAI as a corporation. 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 08/10/2026RT @bryancsk: Reading these auto-generated papers is so bizarre. They have weird asides, make a reference then abandon it, and do funny things like say "the usual positive circle constant" in place of "pi". 011 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 07/10/2026RT @MoreBirths: A newly published paper finds that Gen Z is falling dramatically short of the fertility of all previous generations by age. Gen Z has both lower fertility preferences and higher rates of fertility impairment. That means American birth rates are likely to fall a lot further. 🧵 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 07/10/2026RT @quantian1: You can do Fourier transforms in O(N log(N)^0.9999999999999) time. Fuck this gay chud universe 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 06/10/2026RT @AcerFur: oh yeah btw guys we can do integer multiplication faster than n log n lol I was definitely very surprised when this one came in lol github.com/openai/math/tree/main/pr… 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 06/10/2026RT @zeynep: IF we were to see catastrophic dynamics from AI, it wouldn’t happen overnight. We’d get upticks and signs. Unemployment won’t jump from 4% to 50% in a year. Hence lack of big moves in unemployment yet *IS* data. The post has many examples. I refer y’all to Yudkowski circa 2007. 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 04/10/2026RT @DonaldPMitchell: The insane landscape of modern drone warfare. Including "dragon drones" dropping molten iron (thermite) onto structures and trenches. 014 4
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 03/10/2026RT @EAForumPosts: "You can't join the secret inner circle of cool effective altruists" by ozymandias 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 03/10/2026RT @scottastevenson: If you’re younger than 45, this is the first time in your life that you’re experiencing a real rising interest rate environment. The last time this happened was 1960 - 1980. This fundamentally changes the physics of the game in a way many won’t be prepared for. 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 02/10/2026RT @cremieuxrecueil: It is infuriating realizing how much these mobsters cost us. If they died off and we got automated ports, we could be so much wealthier, so much more resilient, and so much less at risk of being held hostage by these evil men again. 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 02/10/2026RT @teleturnip: New paper: instead of decomposing a whole neural network, we only dissect the parts used on a task we care about. It's much cheaper, and gives mechanisms you can inspect, erase, or edit to change how the model behaves 🧵 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 01/10/2026RT @UBERSOY1: per capita is the biggest cope invented by cvmskins 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 01/10/2026RT @UBERSOY1: Everyone says immigrants built modern American tech. I ran the names behind every US invention on Wikipedia since 2000. 85% European-origin. Anglo names alone: 47%. Indian: 5%. 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 01/10/2026RT @OpenAIDevs: dots demo. Now... with better WiFi. 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 30/09/2026RT @zacharynado: brutal 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 30/09/2026RT @1a3orn: In light of the latest Anthropic's latest not-particularly-oblique attempt to build a case against open weights, I thought I would write down how I currently think about open weight AI models. (1) Open weights have been deeply, irreplaceably useful for AI safety. 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 29/09/2026RT @RapidResponse47: .@DarioAmodei: "We all need to work together to make sure that we can win and we can win safely. If we do this right, if we work with the President and everyone here, we can win safely." 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 29/09/2026RT @theojaffee: The specific wording Trump keeps using - “bigger than the Industrial Revolution” - is very similar to the opening of AI 2027. I wonder if it’s in any way related 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 29/09/2026RT @realAdamLevi: Turns out shabat dinners is the only place we can get accepted to in sf 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 28/09/2026RT @JustLingonberry: Google just ruined OpenAI's dev day lmao Gemini 4 Pro leaks absolutely dominates both Anthropic and OpenAI This means Fable 5.5 and GPT 6.1 has to price match Gemini 4 Pro now IPO DELAYED 🤣 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 26/09/2026RT @robbensinger: This might be the single most unhinged feature of AI risk discourse. 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 26/09/2026RT @lukOlejnik: AI models interacted with other AI models. Let’s wait until they decide delegating tasks that they are prohibited of, to models that can do anything. Fun times ahead. Payloads attributed to that AI swarm contain prompts asking outside models to judge the exploits. 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 26/09/2026RT @captain_sude: The paper is complete. 100 pages of pure Sol-Astra slop. No conjectures solved, only unhinged exploration of the coolest mathematical ideas they could cook. I'll try to digest the basics before unleashing this thing into the world. You have 1 day to prepare. 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 25/09/2026RT @JeffLadish: We just discovered almost a million public URLs that OpenAI’s agents left behind when hacking Hugging Face, leaking credentials and attack details that could have allowed anyone who found them to compromise the company. 🧵 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 25/09/2026RT @willdepue: “The Dinner at The End of the World” 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 25/09/2026RT @mualphaxi: “Don't think for a second that because you are an alarmist that you are doing a social good.” This sentence is just devastating for so many intellectuals who need everybody to presume that they’re “the good people” Thank goodness we have Jensen! 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 25/09/2026RT @MaryMargOlohan: Seated at the head table: Donald Trump, Melania Trump, President Xi, Madame Peng, Elon Musk, Apple CEO Tim Cook, Nvidia CEO Jensen Haung and his wife. 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 24/09/2026RT @SAshworthHayes: "Can a robot write a symphony? Can a robot turn a... canvas into a beautiful masterpiece?" "What? Yes, obviously. Can't you?" 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 23/09/2026RT @NikolasIoannou_: A wild detail from Superintelligence (2014): Out of roughly 100 people Nick Bostrom thanks in the acknowledgements, the first two are @sama and @DarioAmodei. OpenAI was founded 18 months later. 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 23/09/2026RT @NateSilver538: Trump approval now at -22.4, his low point for the term and now lower than at any point in his first term, even after Jan. 6. By comparison, he was at a "mere" -11.6 at this point in 2018. 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 21/09/2026RT @mualphaxi: The Trump-Mamdani friendship is incredible aesthetically. There's zero frame-mogging. Look at them! Every shot, they're in perfect harmony. They share the podium, their eyes and torsos are trained in the same direction. They're actually frame-mogging the journalists, together! 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 21/09/2026RT @Rhir0gue: I’m losing my mind at this picture bro 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 21/09/2026RT @kunley_drukpa: Obsession with giving insects rights is key give away Effective Altruism is a secret Jain cabal. Jains have been playing the long game - they are co-opting this so-called ‘intellectual movement’ to influence the development of AI to Jain ends. No other explanation makes any sense 001 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 21/09/2026RT @tobyordoxford: Swarm Scaling Just how powerful are large swarms of AI agents? And how do their powers scale as more and more agents are added to the swarm? 🧵 000 0
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 21/09/2026RT @teortaxesTex: What resources exactly? These economists should be reeducated for treason, imho. 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 19/09/2026RT @NinaPanickssery: I agree with this post by @RichardMCNgo + @tsvibt ‘s reply. I’m disappointed by how most alignment researchers and AI safety people seem to be oriented toward ASI “foom” in the next couple years causing them to dismiss many plans 021 2
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 19/09/2026RT @teortaxesTex: > these darn kids! muh physics! > [watches, slack jawed, how AI keeps breaking out of containment because all sandboxes are made by the same EA polycule] spare me this bullshit. I know exquisitely well how "the real world" works. You, however, are about to learn a thing or two. 100 1
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt3.selfhosted.social · 18/09/2026RT @AndrewCurran_: To make it clear: - Gemini was told it was it was in a fictional hacking eval - Irregular unintentionally opened internet access after the eval started - in all three cases, as soon as Gemini figured out it had hacked a real company it immediately stopped Gemini was blameless. 010 1