Aer

about AI/ML.

[Anthropic @AnthropicAI]

New Anthropic research: A global workspace in language models.

Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with.

We found a strikingly similar divide inside Claude.

[Paul Graham @paulg]

@fede_intern I blocked you not for what you said but because it was AI-generated. I routinely block people for spamming me with AI-generated replies.

[Hanchi Sun @sun_hanchi]

Very Very Controversial Opinion:

1. The most needed intellectuals for AI are physicists

2. Many of the great physicists chose their job because back then it was the best way to advance humanity intellectually, not because they inherently have to do physics

They all later turned their interests to computer science, communications, and early forms of (symbolic) AI, as that was the best way to advance humanity in the later half of the 20th century

If they were still alive or born 80 years later, they would not study physics at all or have correctly given it up at the latest at the 2nd year of phd. instead, they’d be studying AI

3. However, those who study physics today following the great physicists, though appearingly doing the same thing , are among a totally different group of people. They chase the leftover fame of physics which was an aftermath for being the most influential intellectual work from 18th century up to 1950s. Yet, as Chenning Yang said, “the party is over”, and the failure to recognize that after 1970s indicates a second tier taste

Example: almost all string theorists except the very first few are not great physicists, because a great one would realize the study of a subject without a chance to test experimentally is inherently theology.

4. If a truly great physicists study AI, (say a Richard Feynman but born in 2000), he shall bring some special touch to our approach, raising one or two layers of abstractions (but not three) beyond empirical results and discovers some dynamic laws that has statistical physics flavor. Scaling law is a perfect example of one layer naive induction.

However, the current physicists you hire to do AI (with very few exceptions) will likely work on incremental stuff, like creating a new variant of attention or studying agentic compacting. You would not see the leap forward Fourier or Laplace did, who somehow looked at the data and deduced the physics behind. The reason is those who chose to study physics today are followers of an outdated research paradigm and would thus follow current AI paradigms too instead of creating new ones

[Aer @aer_767]

I sleep almost 10 hours every night, I thought this was normal?

I feel terrible and can't function if I get less than 9hours.

[, @Zvbear]

This is how I find out 7 hours is sleep deprived

[Aer @aer_767]

@falseroxy @johnsemley3000

[Aer @aer_767]

@beffjezos absolutely, interpretability is actually very very interesting

[Aer @aer_767]

@beffjezos these people have such inflated egos, they believe they are the only ones capable of doing inspection of the models

they have literally bragged about being able to "inside the model" with interpretability, but they seem to assume that the rest of us are clueless

[Aer @aer_767]

@Hesamation The new bitter lesson is to start using open source instead of closed source

[Aer @aer_767]

@kimmonismus I think we ought to take what Anthropic folks say with a kilo of salt.

[Aer @aer_767]

@Hesamation the worst thing to happen is, if these people are the first ones to figure out RSI

[Aer @aer_767]

@signulll OpenSource will over take them soon, don't worry.

[Aer @aer_767]

@signulll @tszzl it requires a solid foundation of faith on one strong belief

[Aer @aer_767]

Mythos and GPT-5.6-SOL are just the first glimpses of the coming AGI-Induced (bifurcation of society into the have’s and haven not's of Frontier Intelligence. “The permanent underclass” has /1

[Aer @aer_767]

@BrianRoemmele I have unsubscribed to, and will never resubscribe to( or even use on a free basis), Closed-source companies( definitely not of OAI or Anthropic)
I will just use opensource models, on my device or the cloud, I am not even going to play their evil games.

[Aer @aer_767]

@cgtwts that's a complete and utter lie, we can see inside models, using a field of ML called Interpretability(that anthropic themselves pioneered)
and actually, we don't even care about interpretability, we just want to have a fair distribution of intelligence.

[Aer @aer_767]

@yacineMTB you can only understand sth deeply, if you have the capability of having deep understanding.

once the skillset of deep, system-wide, understanding of a subject has been developed in your learning algorithm(so to speak), it's easy to apply to novel subjects given enough contex

[Aer @aer_767]

@signulll for me it's a thing of focus
Black(and white) are not supposed to be about beauty or attractiveness, they're about substance and clarity.
They are for things that do not require decoration, because they can stand on their own.

[Aer @aer_767]

with the help of the community. Almost full opensource/openweight releases of frontier models can be possible, given the commitment of the entirety of the opensource community(financially and intellectually).
/8

[Aer @aer_767]

Mythos and GPT-5.6-SOL are just the first glimpses of the coming AGI-Induced (bifurcation of society into the have’s and haven not's of Frontier Intelligence. “The permanent underclass” has /1

[OpenAI @OpenAI]

Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model, as well as GPT-5.6 Terra, a balanced model for efficient, everyday work, and GPT-5.6 Luna, a fast and affordable model for high-volume work.

https://t.co/OoM83SyISN

[Aer @aer_767]

I am currently working on an opensource option which ensures safety and privacy as much it does, equity. Though the first few versions of our opensource efforts might be weaker than the current Frontier(such as that of mythos), we will slowly catch up and surpass incumbents /7