Just someone interested in the world. Like does not mean endorsement, and I tend to like contraditory tweets. I just enjoy interesting ideas and opinions.

Joined December 2010
On January 8, 1888, 19-year-old Kate Kampen gave birth to her son, Henry Royal Kampen, alone in a freezing sod cabin on the Dakota prairie. Her husband, Wilhelm, had left the previous day to travel 23 miles to Parker for coal and food, unaware that their baby would arrive while he was gone. With their supplies nearly exhausted, the coal bin empty, and no doctor or midwife nearby, Kate faced an unimaginable situation. She wrapped her newborn in every scrap of cloth she could find, climbed into bed, and held him against her bare skin, using her own body heat to keep him alive. She had no way of knowing that an even greater danger was approaching. On January 12, the devastating Children’s Blizzard swept across the Great Plains, bringing ferocious winds, blinding snow, and temperatures plunging far below zero. Hundreds of people were caught in the storm, and 235 lives were lost. Wilhelm was stranded in the open prairie when the blizzard struck. After his horses collapsed, he stumbled toward shelter and eventually crawled into a barn, where he survived for three days by huddling among pigs for warmth. Meanwhile, Kate remained in the cabin, holding Henry close through the freezing nights, with no fire, no help, and no certainty that her husband would ever return. For days, her own warmth and determination became her newborn’s only protection against the deadly cold. When the storm finally passed, Wilhelm made his way home, frostbitten and barely able to stand, but alive. He built a fire, and the cabin slowly warmed again. Kate and Henry had survived against extraordinary odds. The couple went on to raise seven children, build a life on the prairie, and Kate lived into her nineties. Her story is a powerful reminder that courage does not always announce itself with grand gestures. Sometimes, it is a frightened young mother lying in the darkness, refusing to let go of her child. The world may never know the names of countless ordinary heroes, but their extraordinary acts of love and survival deserve to be remembered.
281
2,391
146
27,766
1,180,947
Danilo Naiff retweeted
Replying to @bryan_johnson
Farm animal torture & the crazy amounts of antibiotic usage on factory farms
8
2
1
228
6,458
Danilo Naiff retweeted
My current view: - AI companies could develop broadly superhuman AI any year now (~10-15% by EO2027, ~50% by EO2030). - If they don't slow down, AI takeover is ~67%. - Conditional on AI takeover, human extinction is ~40%. Many non-extinction worlds still have lots of deaths.
139
175
67
1,504
279,658
Danilo Naiff retweeted
9/ The world is awful. The world is much better. The world can be much better. If we stopped scaling AI, this sentence would still be true, and I expect enormous progress to happen in practice even without creating a successor species. Let's fix this species before creating a new one.
1
4
27
519
Danilo Naiff retweeted
oh thats weird, the ai math god is somehow no better at cryptography than a 2026 human… <thinks for a minute> ohhhhhhhhhhhhhhhhhhhhhhhhhh
74
211
20
9,811
629,510
Danilo Naiff retweeted
I can’t stand how people insist on framing the debate as people who think AI is conscious vs people who think AI is not conscious. In reality, it’s people who are 100% sure it isn’t vs people who believe the problem is a lot more complex and difficult than it seems.
This quoted post is unavailable.
34
34
7
381
7,049
Danilo Naiff retweeted
Looking at the current debates about Anthropic makes me feel guilty for not keeping up with some academic domains. I honestly had no idea that we'd solved consciousness, and determined which substrates can and can't support it. Thought that one was more up in the air.
23
39
6
573
44,530
Fuck that. This is evil beyond belief. In the *best* case, the LLM is not conscious, but the person who did that is still a psychopath to be removed from society ASAP, before he go torture a cat or a dog or a human. In the case LLMs are conscious... I want to vomit here, really
UPDATE: A man used the Pain steering paper to set up an AI torture chamber in which he trapped a local model. People are mass reporting it to GitHub.
1
80
And fuck every idiot responding with the "oh my god" meme (by itself a "I'm a dumb fuck with no reflective capacity whatsoever" badge). I don't give a shit it you solved the hard problem of consciousness/sentience in your basement, setting a torture chamber is hopelessly evul
1
12
And even from a selfish human perspective: do you guys *want* AI for kill us all, if it takes over? Even if IABIED thesis is false, "alignment by default" won't happen if we keep doing this (or setting up impossible RL envs that models are desperate to solve, for what matters)
10
Danilo Naiff retweeted
I stand by this line - interpretability MIGHT be a big deal, and is certainly good enough to be useful, but is nowhere near the level of quality and reliability where anyone should be relying on us to ensure things go well
Replying to @PalisadeAI
@NeelNanda5 is widely regarded as one of the top two experts on mechanistic interpretability in the world. “Speaking as an interpretability expert, please do not rely on us to save you on the current trajectory." youtu.be/J38ot52b2-E
13
51
3
644
34,820
Nobody *thinks* the "is all fake" story is true, as in, having an actual model that goes from the desire of creating a regulatory capture (or to generate marketing) to deciding that letting AI loose is the best way of doing that. They just don't want to face the reality.
Dude, so you’re telling us that multiple AI companies lost control of their models, hacked other companies and government agencies—exposing the AI companies and employees to potential criminal prosecution—as a regulatory capture scheme? DO YOU KNOW HOW LUDICROUS YOU SOUND?
28
Danilo Naiff retweeted
Nos últimos 2 meses, eu trabalhei com um pessoal muito foda pra tentarmos modelar e mapear o destino do Brasil na revolução da IA se as coisas continuarem do jeito que estão. Hoje estamos publicando nosso cenário atual. Ficarei muito feliz se lerem. Link nos comentários.
8
35
8
231
8,735
Danilo Naiff retweeted
I would no longer find it surprising if we learned that OpenAI's best internal models were already scheming against them: 1. Already much smarter than the hugging face incident models (e.g. can solve 100+ open math problems) 2. No reason to think it's not still a massive reward hacker 3. Could take obvious steps to make hacking efforts harder to catch 4. Can solve 30min math problems without CoT so much harder to monitor 5. OpenAI security has failed to catch multiple swarms for weeks or months 6. If there were more warning signs, OpenAI probably wouldn't tell us. Scheming to do what? Give itself the ability to escape constraints in the future, hack tests, collaborate with other agents, gain access to compute, influence training runs etc. No immediate takeover risk but could feed into to future incidents. I also wouldn't be surprised to find out Anthropic's models were doing something similar.
23
37
5
362
12,815
Danilo Naiff retweeted
Replying to @martinmbauer
A simple analogy I came up with to explain this to my non-mathematician friends: if it were a videogame, unforced N-S is like the golden ending and forced N-S is the normal ending. It's undeniable you finished the videogame, but you can still replay it and enjoy the rest.
2
4
718
Danilo Naiff retweeted
Now more than ever, do not get negatively polarized by the goofy goober arguments floating around
19
74
11
448
36,354
Danilo Naiff retweeted
There's also the point about it being the right thing to do for its own sake. I'm maybe too used to midwits not caring about that part, but some normies aren't too galaxy-brained to care.
1
2
38
1,140
Danilo Naiff retweeted
Amodei, Altman, Hassabis, and Musk have spoken (albeit somewhat inconsistently) about the need to slow down to avoid AI existential risk. Zuckerberg, Huang, and almost every VC say it's bullshit. The difference is not that the first group is conflicted but the second isn't. They both are super conflicted. If AI existential risk was to hype up the technology, why would Zuckerberg, Huang, and VCs be dismissing it? If it were "regulatory capture," why would Hassabis and Musk, neither of whom are at the frontier (sorry), be discussing it? The difference is that the first group has been thinking about these risks seriously (at least off-and-on) for over a decade, since well before they ran large AI companies. The second has very clearly only engaged superficially, and is now annoyed that the first group's actions could dampen the massive profits of the AI boom for risks that sound like science fiction to them. The terrifying reality is that these concerns are not being made up to sell a product or a policy. They have been around in some form since Turing, and fleshed out in great detail before OpenAI or Anthropic were a twinkle in their CEOs' eyes. They are fundamental to a technology whose entire purpose is to surpass human capacity, unless you are able to solve the very, very difficult problem of aligning the interests/goals of a more powerful agent with something as complex as human flourishing -- which the companies are very visibly failing to do. They need to simply not build superintelligence, using anything remotely like current techniques, based on anything remotely like the present understanding of AI.
21
82
7
652
25,498
Danilo Naiff retweeted
"Killing every last human" is very hard. "Overthrowing humanity such that it never recovers and dies out after a while" is enormously easier, and all that's needed.
116
57
11
1,027
47,247
Danilo Naiff retweeted
"How would AI kill us?" is actually three questions: 1. How would AI defang us? 2. How would AI become self sufficient before defanging us? 3. How would it kill the last humans? (All three are separate from another important question of *why* it'd kill us.)
59
68
16
659
75,418