upvote
> After brooding on this news all day, I finally laughed and told myself: “I should just vacuum my apartment. It is messy and needs cleaning. What else can one do? ”

Smarter people than me called this the normalcy overhang and it feels the erosion is becoming deeper by the day.

reply
... How was the proof of the GRH, that mathematicians already assumed to be true anyway, supposed to change anything? Did we learn anything new? Did we unlearn something false?

Even the four-colors-map theorem had more actual impact.

Congratulations to Alpoge (and maybe Massulo), his teachers, his colleagues, his predecessors and everyone who made the tools he used. But nothing changed much.

reply
We learnt that it's true and AIs with best humans could solve a very hard problem no human could solve before. This implication is probably more consequential, do we know think humans and AIs can solve the problem of developing an AI capable of solving riemann hard problems by itself?
reply
> I thought I could no longer be surprised by AI progress. Yet I had not truly believed that Claude could prove a conjecture as challenging as the Riemann Hypothesis.

I don't really get this. I mean I am not surprised by AI progress but this means that I expect AI to do at least anything I imagine a human could do, including proving RH.

reply
I am a layperson nobody and never understood this existential dread about AI until now. I am not even worried about RSI.

Even if Anthropic has already hit the limit on capability I do not think that this unfathomably valuable math machine should be in the hands of a private company. They become the arbiters of all or certain areas of intellectual progress or knowledge or whatever you want to call it?

Until today I just thought these models were weights and compute. But to produce something like this there must be a complex system that is hard and expensive to replicate. I feel crazy for even thinking this but I am now thinking seriously about what nationalization means. Not that I have any control over it, but I think it might be necessary. Me two days ago would punch me.

reply
If one company did it, others will do it as well. There is no moat.
reply
Capital is the moat, because big data and compute are what unlocks these results.
reply
Capital is not a moat though. Any other company can raise comparable capital, if the investors believe it will pay off.
reply
You would prefer this tech be in the hands of the Trump government?
reply
No. Quite the problem isn’t it? Even if it ever did happen it is not happening that fast. But the analogy kicking around in my head is we already would never accept a private company with nuclear weapons without government involvement right?
reply
Yes I would rather nukes be in private hands than in Trumps personally.
reply
Not in the hand of the government, but of the people.
reply
The people is a rather abstract term. Nobody really knows what "the people" want. Many things "the people" want alienate about half of the people for example.
reply
"The People" is not some physical coherent entity that can hold something in its collective hand. It's an abstraction. How do give something to an abstraction? It's like saying I want to give an apple to pi.

That's why governments are necessary. And, in this case, I don't see how the Trump administration running the tech would be an improvement.

reply
Well, yes. I can vote them out. I can't vote the ANT leadership out if I don't like what they're doing.
reply
Thats theory, anthropic can live or die by the market. The usa government is effectively immortal and can murder you legally.
reply
And the mathematicians in existential crisis plot continues. Deepens.[1]

It could prove all the theorems in the world. At this point it/they are running on electrical appliances that you can thwart by turning off the power. What is actually concerning are the humans behind these enterprises. It’s all just the same sociopathic thirst for absolute domination. Same old “human nature” as the last two hundred thousand years.

But Scott Aaronson’s smartest people he knows is concerned about computers ending the world. Not the humans who are making these applicances doing the same. I guess intelligence is a bit more compartmentalized than the smartest people would like to think.

And thus I paced around my office, thinking about the last five Hacker News submissions on maths, but more organically presented as if I ran into them on the street. Will computahs end dah world? Then the story circles back to whatever mundane thing I was going to do next. Ha ha: There’s nothing that I can do with humans wanting to dominate the world, sure isn’t.

[1] https://news.ycombinator.com/item?id=50489633

reply