AlexCoventry

> Terence Tao is arguing

It's a guest post by Amit Sahai, FWIW.

IMO, it's almost certainly being abused. I just assume that all US domestic communications are being recorded whenever possible.

It can be a textbook designation and a capricious punishment at the same time. It's almost comical, to suggest that anything about this decision was reasonable. Makes me long for the days of n-gate.com.

IME, it's best to ask an AI to criticize your work and suggest changes without providing actual rewrites, and make the changes you agree with in your own voice.

To all appearances, OpenAI just solved [1] a Nobel-level (Fields medal-leval, strictly speaking) math problem, but you still think it's hype?

[1] https://openai.com/index/navier-stokes-solution/

I think the idea is that AI itself is going to massively accelerate its own development, and AI with real-world competence is coming very soon. At the rate things are going, it wouldn't suprise me at all if we had mass production of AI-designed systems in the next six months, actually.

I don't think we should assume a millenium puzzle has been solved, yet. Astra showed impressive capacity for cheating when it was faced with impossible cybersecurity challenges. It seems equally plausible at this stage that it's found a bug in Lean.

I'm not a lawyer, but I think the comparison to copyright law is invalid. Such a dispute would be governed by contract law.

Only if you have ready access to the hardware to run the AI models.

Which "elaborate guerrilla advertising campaigns to drive up capital funding interest" were you referring to, in regard to Chinese models not needing to set up message boards for inter-model communication?

You think OpenAI framed themselves for legal vulnerability to Huggingface?

I'd be interested to read more about that.

As far as I can tell, IBM has mostly been putting out junk for twenty years.

Yes.

I think the most likely way for Zitron to find vindication is through the world war we seem to be shambling into. That could easily put a stop to AI improvements, because the supply chain is so fragile. But that would have nothing to do with his ostensible reasoning.

> The issue is that the process of "testing" was careless.

Yeah, who thought that giving agents with this much capability any internet access was a good idea? I'm not a Yudkowskyite, but surely entirely in-house, offline infrastructure is table stakes for AI containment.

Dispassionate search for and representation of truth?

> Additionally, they've redefined that a pandemic is (removing severity from the definition)

Over a million Americans died, but OK, that's not a severe problem...

The Economist had a similar article [1] about Han chauvinism leading to rejection of the Qing dynasty, and causing quite a headache for the CCP.

> In China’s official narrative, the country suffered a “century of humiliation” of foreign invasions, starting with the first opium war in 1840 and ending only with Mao’s rise to power in 1949. But the 1644 view—cheekily expressed by some online as “three centuries of humiliation”—argues that subjugation began with the Qing. According to this school, the Qing were anti-Han interlopers and, making matters worse, feckless in repelling foreign forces.

> Dredging up all this history is uncomfortable for the party. The territorial boundaries claimed by modern China—including Tibet, Xinjiang, Inner Mongolia and Taiwan—were brought into the Chinese empire by the Qing, not the Ming. Grievances about the Manchus also cut against the ideal of ethnic harmony advanced by the party. Although much of what forms modern Chinese identity is derived from Han traditions, it is unseemly to talk openly about the dominance of this one ethnic group, which accounts for 90% of the Chinese population. The party looks askance at “Han chauvinism”, believing, with reason, that it may lead to more resentment from non-Han groups.

[1]: https://www.economist.com/china/2026/07/20/why-criticism-of-...

I think it's good that someone is making this point, anyway. For sandboxing a super-capable offensive-security AI, you would think that cloning PyPI and running it as an offline service ought to be table stakes, but apparently that's not how OpenAI saw it, for instance.

Time frame is '21-'24 (Biden term) and from U Chicago. Pretty sure this is right-wing propaganda.

I suspect Hegseth's atavistic attempt to turn the US military back toward a culture of machismo is resulting in misery for anyone who doesn't fit the part, whether they're skilled at their duties or not.

I would be delighted to see a glut of compute. I'm not optimistic, though.

SpaceX mainly showed how corrupt securities regulation has become in the US. It was already quite corrupt, but there's no way that IPO would have flown under the Biden administration.

I think there's not much to model. Either AI actually becomes a hot commodity, or the people invested in these deals are going to take a bath.

Power is more decentralized in the US, even now with Trump.

Do you doubt that Xi Jinping is a Marxist? https://www.youtube.com/watch?v=DprKDXRlubw

What Xi Jinping Wants: https://www.youtube.com/watch?v=DprKDXRlubw

At least they can resist it, as we saw with Anthropic.

Open Source won out because the cost of compute fell through the floor. I'm not sure whether we're going to see a similar dynamic play out this time, although I would greatly prefer it to.