With SO the experience always felt like you were being chastised for everything you did. It was quite off putting. AI coming along and just giving you working solutions without any attitude was just a more pleasant way to work.

So why are these models more successful than mathematicians at solving these problems? Is it because of the computational power they bring to the problem? Is it because they use new techniques not thought of before? Or is it because human mathematicians have been approaching these problems in the wrong way?

A bit part of it is that no one human mathematical genius can hold all of math in their mind at once. The surface area is enormous. Humans specialize. The model has the luxury of finding obvious interactions in widely disconnected fields that no human would have found.

This seems like a like a huge advantage, the models can view everything. I wonder if they're being used to look at problems in physics like quantum gravity.

I am seeing a lot of high energy physics preprints on arxiv coming with AI acknowledgement recently https://arxiv.org/abs/2610.10673 https://arxiv.org/abs/2610.09093

But they aren't more successful. They can solve this because of standing on a huge body of work made by working mathematicians. Without it AI would get nowhere. Now AI are burning the minds that fed it for short term gain.

When AI has exhausted the existing math literature (and found a bunch of errors, I'm sure) it will have to start producing its own conjectures. Millions of them, even billions. And using all the experience from trying to prove them to improve.

You're acting like math is some sort of infinite game of automating unit tests. There is no shortage of conjectures and generating them is cheap. The struggle is finding the ones that are actually interesting. They only have value when they can unify disparate domains or reveal something that was previously hidden, or (the thing that I see most Mathematicians lamenting today) if they force the creation of new abstractions to solve.

Insight in mathematics doesn't emerge from brute-forcing conjectures. There is a recognized aesthetic economy that exists and without a semantic filter to detect the significance in any of these proofs we just end up with a lot of noise.

Oh, I'm sure most of those millions or billions of automated theorems will be uninteresting to us. But they will, in aggregate, be useful for the AIs, in improving the models and identifying lemmas and abstractions.

Yes, it would be very much like automatically generated tests. These are not worth much by themselves, but in massive bulk they can do interesting things.

Why on earth would open AI pay for that though? There is no value in it anymore. The pointlessness of this destruction of math really is depressing to me.

They would do it if it would improve the general intelligence of their model(s).

This is giving me Juicero vibes.

I think open source code would sit in the opposite end of the spectrum to Juicero.

Better to site the primary source, its like googling something and then siting google as your source.

Let the enshittification commence.

We know what this is really about, its to give the UK intelligence services the ability to spy on US citizens and then pass the information on to US intelligence. We'll probably get similar laws in all the other five eyes as well.

That's not what this is about at all. Its British NGOs lobbying US states to pass state-level digital-ID laws. The end goal seems to be to make anonymous posting on the internet illegal. Its more about mass censorship than surveillance.

Mass censorship and surveillance go hand in hand - you use one to enforce the other.

Isn't that what mossad is for?

Just charge me more, don't cut the size down. It's like I'm buying bite sized chocolate bars these days and the price is higher. If I'm going to pay more give me the original sizes.

I prefer smaller portions. If you want more, you can always buy two.

I am guessing there is some kind of political incentive to create stealth inflation (though it's literally 'deflation'), which is largely unrecorded, vs. traditional inflation, which is a vote-loser.

Because it’s not only a social media company.

Small retail investors comprise 66% of the market in SK. It's usually single unmarried people who make ridiculously big bets to try and get enough money to buy a house or get married. It's something to be pitied not hated.

Okay yeah that's very sad. People being tricked / kind of forced to bet money they can't afford to lose is very shitty.

The Numbers is such a brilliant site, this explains why it came back in such a stripped down form. AI and prediction markets giving me more reason to hate them.