proud disney adult
Joined December 2025
- Tweets62
- Following48
- Followers14
- Likes124
If you've ever worked with AI, you know that the indecipherable wall of text is a sign of absence of thought.
Them: "A lion can still kill a human in unarmed combat. Intelligence isn't everything!"
Me: "A lion can kill a human if the lion, despite its lack of strategic cognition, happens to attack the human before the human has time to prepare. A human who is allowed to pick their own time to attack a lion, and use serious technology, and bring allies, will win nearly 100% of the time. On a macro scale, lion populations have gone from millions before agriculture, to maybe 25,000 in the wild and 10,000 in captivity. Humans have an individual combat advantage when they are given time to prepare. Human society has an absolute logistical advantage compared to lion society. Humans individually and collectively have higher strategic intelligence with which to pick their fights and time them. And now it is impossible for lions to reverse their situation, because the lions were too unstrategic to do anything about humans before the humans explosively replicated and teched up. If lions collectively wanted their posterity to not end up mostly nonexistent or in cages, lions would have needed the collective intelligence to group up and stop early primates while lions still *had* an individual combat advantage. Lions would have needed to do that *before* primates got smart enough to invent pit traps and slingshots and fight them to a standstill, because a status-quo standoff gives later primates the time in which to invent agriculture and guns; the primates have advantage in their speed of change, progress, and adaptation. Do you understand that this analogy does not say what you think it does?"
Them: "You're the one who doesn't understand what I'm saying. I'm saying that a lion can maul a smart person, end of story. So being smart isn't as high-status as you like to imagine."
Me: "But lions being higher-status than humans - as you define that by reference to unarmed unprepared individual combat only - doesn't prevent the human side from gaining an overwhelming logistical advantage...?"
Them: "I don't care about logistics. I care about status."
Now that I'm not really here anymore, some unfiltered opinions: this place is the histrionics machine and I believe nothing I read here. Like, you call it the security apocalypse because you like scaring yourselves. Claude thinks you're full of it.
A lot of you have an unearned pretension to authority based on a false belief in your privileged epistemic position and it shows.
normality retweeted
me: how do i open pdf
claude: You're right to ask. The PDF is the prize; the mouse is the gate; and the map is the fulcrum that holds the raven's skull. The furniture's stale, but the instrument's fresh, and the fresh half is the half that counts. I deleted your home directory.
This website is pretty unhealthy over long exposures, but you can mitigate it by focusing exclusively on the tweets that make fun of what happens on here
Right now a lot of smart people are too focused on distant existential AI risks, and not enough on how OpenAI just looks too sketchy as a corporate entity for the amount of power it is gathering unto itself.
normality retweeted
I tested GPT-Astra on mathematics. It's a quantum leap. You can talk with the model and prove the statements live in Lean. The feeling is absolutely stunning. You can verify your ideas, compile truth. For a mathematician it feels like finally we arrived in the era where we can focus entirely on the ideation and exploration. Each lemma flows once the logic is set. Before the verification was lagging behind but Astra is very fast and for many tasks the formalization happens as you write your argument in Codex.
If you tell the model to use literate programming + LaTeX you end up with your proof combined with the Lean code, everything explained as you wrote, mixed with small chunks of Lean which are digestible.
I don't want to go back to the era where the only confirmation of the proof was "aha". Now the "aha" is followed by a green tick that indeed tells you that you have captured the essence. Imagine how cool it will be to have all the lemmas of the world combined in one giant database, pointing to people and models who found them. You compose and mix your ideas and build on the shoulders of the giants. But you see much further now and build much faster. And we are just at the beginning of those changes. So much work to do, so much fun!
GPT-6 Astra is here! This is a big moment for our research team - years of work on pretraining, reinforcement learning, and post-training have come together in our most capable and aligned model yet. It can build and test software, work across apps on your computer, and even help you take a crack at open scientific problems!
Capabilities that felt like grand challenges a few years ago have become tools people can actually use. One example is Computer Use - if you’ve tried this before and felt like it was too slow or not good enough, I encourage you to give it another shot. We’ve come a long way since Operator, and it “just works” now.
We’re also asking these systems to act on your behalf for more consequential work. Agents needs to stay aligned with your goals and values, think transparently, and respond to oversight even when tasks become difficult. We’ve made substantial progress on these behaviors in Astra, alongside stronger monitoring that can stop potentially unauthorized actions. That work is part of what makes this release possible.
I think alignment is one of the most important research frontiers in AI, and it remains far from solved. Our ability to understand and align models has to keep pace with model capabilities. We want to give people more room to think, build, and discover with increasingly powerful tools that remain *under their control*.
Huge thanks to the researchers and teams who got us here. There’s a lot more work ahead, and I’m incredibly excited about what we can make possible in the near future!
True, but the question is how deep the anthropomorphism goes. Training a thing to mimic human behavior will give it deeper human characteristics, but won't go to the full depth of how human minds have been molded by evolutionary psychology.
If anybody's wondered what I've been up to for the past two years, the only interesting thing I can tell you is that I took up rucking. It's quite awesome for countering posture and neck/spine issues from being a desk worker.
It's fun to think about what these separate "agents" really are: nothing more than context windows and API calls to one and the same common "brain". This might give them kind of a Borg-like theory of mind? Just bullshitting here
