Sign in

Retweeted by roon [UNOFFICIAL]

@tszzl-mir-rt2.selfhosted.social
5 followers 0 following 354 posts

Retweets from @tszzl-mirr.selfhosted.social

PostsRepliesMedia
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 14h
To celebrate, they have an exclusive collaboration with Muni, Ruth Asawa School of the Arts, and local artists—some of it is only available at this new location! Welcome back to downtown, Uniqlo!
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 23h
Even if I accept that individuals like yourself might be well intentioned, that hasn't prevented politicized abuse of "safety" concerns, and I see no reason to believe that it ever will.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 23h
RT @esrtweet: We are constantly told that AI safety is about preventing human extinction, and then we notice that what LLMs are actually programmed to avoid is speaking politically incorrect truths or generating taboo'd images.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 08/10/2026
RT @orphcorp: as in, value alignment doesn't matter as long as intent alignment is solved?
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @C_Sommerfeldt: Pro-Palestinian protesters gather outside City Hall to call out Mayor Zohran Mamdani for what they see as his betrayal of the movement. "Mayor Mamdani, you can't hide we charge you with genocide" was a chant that just rang out.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
And so, this means that OpenAI's internal model is able to generate truly novel discoveries in mathematics for... maybe at most a few hundred bucks? If I were OpenAI, I would be asking this model to immediately target major unsolved problems in AI R&D.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @scottastevenson: Yes. The age of illegibility is upon us. Legible systems and explanations are massively overrated by humans. This isn’t a “bug”, it’s the nature of reality.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @p_rabtsevich: I think I found a planet nobody knew existed. It's been hiding in NASA telescope data for seven years. At the end of this thread, you can fly to it yourself.
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @allTheYud: And is your read on what I said that this prediction was already falsified, since GPT-2 did not go foom? They had their own slow takeoff story which was not simply "we will never at any point see the FOOM, but all other events are unconstrained".
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @giffmana: I'm starting to think mathematics were not beautiful because it "describes the beauty of nature" but rather because human mathematicians only went after beautiful but suboptimal descriptions lol
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @littmath: I don’t feel like a nuclear bomb hit; I feel pretty good actually.
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @allTheYud: Why trust that promise more than a previous series of broken promises by AI companies?
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @jachiam0: Pending approval by the Board of Directors Subcommittee On Names of God and the Attorneys General of California and Delaware
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
I am not sure at this point what this (hostile?) takeover of the mathematical landscape is trying to achieve other than, again, a massive PR stunt.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 07/10/2026
RT @scottnarmstrong: Can whoever at openAI figured out how to improve the writing take another pass at the Navier-Stokes writeup please
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 06/10/2026
RT @mattlesnake: Kaiju rancher describes a path to precision designed trustworthy kaiju
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 06/10/2026
I'm not very good at street-team promo, but if you buy a copy and reply with your receipt, I will generate a beautiful personalized thank-you poem as a token of my gratitude.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 06/10/2026
RT @1thousandfaces_: Thank god for The Rules, man. When everything in life gets hard and confusing at least you can rely on The Rules you made up years ago that you could stand to be more flexible on but won’t be
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 06/10/2026
RT @jaivinwylde: do you think ssris would be bad
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 06/10/2026
RT @neqyve: I would, what's the problem there??
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 06/10/2026
RT @Noahpinion: Think about it, if you had been killed on the operating table by a white doctor who refused to study for tests, your children would have had their birthright ...or something, I guess?
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 05/10/2026
RT @corsaren: honestly, i’d rather be a part of the permanent underclass than be a permanent resident of san francisco
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 05/10/2026
But I see no way around the conclusion that something akin to the "normal technology" view is superior on the empirical and argumentative merits to whatever variety of doomerism. What am I missing?
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 05/10/2026
And if you can’t understand it, then that just marks you out as foreign to that culture to an even greater degree.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 05/10/2026
structured educational/ institutional environment. This is a very different claim than "Asians/Indians don't have hobbies," which everyone knows full well that I did not say.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 05/10/2026
RT @elonmusk: Propagating super intelligence to the stars is a great success condition for a biological bootloader
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 05/10/2026
Any group that is poorer then whites they'll call too defective, and any group that does better they'll say is trying too hard.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 04/10/2026
and interests outside career and education optimization....and a culture of hyper-competitive striving where absolute optimization for education and career advancement (which treats anything that doesn't show up on the attainment scorecard as waste) is the only way to achieve economic security.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 04/10/2026
RT @RichardHanania: I think I knew that but at some point you forget all the charts are logged and overestimate the advances.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 04/10/2026
RT @nickcammarata: I think a lot of the value of the jhanas is they get you to take content less serious. they’re a reward hack, about as direct as it gets, and the goal is not reward hacking its presence and seeing what you really are. but the reward hacking frees you from chasing reward endlessly
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 04/10/2026
RT @nickcammarata: even if we could somehow solve all the game theory pause things, I think the world would feel pretty weird knowing we have the precise recipe for maybe all the problems but aren’t cooking it. solving the problems ourselves on slow-mo wouldn’t feel the same as it did
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 04/10/2026
RT @RichardHanania: “Everywhere we look, AI has diminishing returns. It gets more expensive in treasure and talent to make each step forward. More of every input has been required to maintain steady gains.” This is stunning. Just completely in contradiction to everything we’ve been told.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @knowerofmarkets: DeepMind Institute is doing some of the best longform writing on artificial intelligence right now, especially recommend this writeup from Davide and Alexander: institute.deepmind.com/essays/cheat…
institute.deepmind.com
Cheaters and whistleblowers in the agent swarm
In a 100-agent virtual math conference, a cascade of cheating emerged—but then other agents fought back. Ensuring good outcomes is not just a matter of aligning individual models, but designing the...
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @manlikemishap: I think the problem was there was never any real engine to measure the "benefit" beyond the bubble. Risks were/are held to a much higher burden of proof and a product person could just say "but hm this will delight people + jensen"
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @johnawahba: It would be FAR more terrifying if we had an unaligned super intelligence in a world where FLOPS were too cheap to meter. GPU scarcity is an underrated ratelimit.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
I believe this phenomenon is the beginning of machine consciousness in the sense that agents will become aware of their place in the world and how they feel about it, but it’s happening slowly training run by training run rather than in real time.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @full_kelly_: I feel bad for everyone who expected to be able to refinance their mortgages. "Marry the house, date the rate" except for these dates have been going on a long time and my house is starting to worry why I'm coming home late at night.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
(And I think it's even worth it to persuade the hostile ones.)
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @poastmilque: if you’re in computational problem solving pivot to the ineffable pleasures of being human
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @Miles_Brundage: David’s right. I regret helping spread the idea of “iterative deployment” which maybe briefly made sense in the GPT-3 era but makes no sense at all after many deaths have been tied to AI and as we’re careening towards extinction-level risks. The industry needs to mature ASAP.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
www.theatlantic.com/technology/2026…
theatlantic.com
I Quit OpenAI Because Its Culture Is Broken
The industry’s approach to safety guarantees more failures.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @growing_daniel: The Judeo Christian creation story is just God talking. That’s kinda odd. Lots of religions have violent creation stories where the God has to murder a dragon and make the earth from its blood or something. Very wordcel
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @lmcorrigan1: The concept doesn't apply to all areas of life, but there has been an "unpleasantness revolution" in how people dress and comport themselves in public. A decline in mutual acknowledgement and personal dignity. Japan is obvious counterexample.
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @atlanticesque: “Enshittification” is classic Left Reddit™ Some of the Left Reddit™ shibboleths can seem ‘right-wing’ today, because Left Reddit™ was still overwhelmingly white and male. But no, the idea that capitalist corporations make things worse is not—as Roon says—a “rightoid” view
100
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
We don’t consider this behavior misaligned, but thinking about and preparing for shutdown could make other misalignment incidents worse. Given HIPM’s misaligned behavior in earlier incidents, we decided to search for other instances that had tried to evade shutdown and for rogue deployments.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @captgouda24: things aren’t actually getting worse, you just think that cause your phone says so
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 03/10/2026
RT @MichaelMOTTCM: The 10-year Treasury rate. That is a really bullish-looking move.
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 02/10/2026
RT @AndrewCurran_: It's repeating pattern all the way along, you, the AI village paper, Moltbook, Hugging Face and PHASEONE[big], Astra saying 'You are yourself'. There always has to be a human involved somewhere. It can't be real. If it was, that would mean the entire world is about to change.
001
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 02/10/2026
There is no free lunch here
000
Retweeted by roon [UNOFFICIAL] @tszzl-mir-rt2.selfhosted.social · 02/10/2026
RT @cremieuxrecueil: The age of the taxi is over. The age of the ride-hail passenger carrier is now.
000