> You'd be laughed at if you said that ChatGPT could help you with graduate level mathematics in 2024, but this year, AI models on simple prompts are solving previously unsolved Erdos problems.

I'm curious, do you have a graduate degree in mathematics?

I have a graduate degree in mathematics. AI models can absolutely help you do research math in 2026. I recently asked chatGPT to prove a result which I know to have been published in Advances in Mathematics (a pretty good, but not top tier, journal) this year, and it gave a correct proof which was completely distinct from the one that was published.

It's a press release.

They were not asking for your individual opinion. They were asking why any given person would need it.

It turns out I am any given person

Yep... I've met him and I can confirm that he is in fact, any given person. To that, I might add that I'm in the same situation. AI is a convenience and light entertainment, and as such, I'm willing and ready to pay exactly 0 USD for it privately and in my business. If the price rises beyond this point, they for sure have to move it from convenience to essential and so far, I have not seen any shadow of essential.

You are missing the forest for the trees.

Just consider the math there for a second, when you factor in the average US wage.

But we know global GDP per capita (a proxy to wages)- the US represent about 25% of total global GDP (a metric which accounts for US wages being higher than average). I’m not being contrarian, I genuinely think the addressable market is the global market and not just the US (and by a wide margin) and as such thats the real potential of anthropic/openai/et al.

You're partly right. But OTOH, China is (pretty successfully) developing its own AI solutions, and even the (former?) US allies in Europe, America and Asia have become painfully aware that they are dependent on a hostile US administration and tech companies cozying up to that administration, and will be wary of further deepening this dependency, so they will also prefer home-grown solutions. So the addressable market for US companies is much smaller than the global market, even in countries that could theoretically afford Anthropic's and OpenAI's prices.

Globally available AI means cheap LLM pilots offshore means huge wage deflation in white collar jobs in the US.

…assuming the AI pace of progress slows down enough that there are still meaningful white collar jobs to have.

This is just the offshoring and remote discourse all over again. It turns out that the prestige of having a big office full of workers that the CEO can see is well worth the massive expense of siting it in California and paying California salaries. For whatever reason.

(also I suspect the anti-globalization discourse will get even more pointed)

> There is no evidence yet in the labour-market data of AI destroying many jobs.

Literally from the article https://www.economist.com/leaders/2026/05/14/prepare-for-an-...

Surely you're not referring to this prediction that was so bad, it's now a meme: https://news.ycombinator.com/item?id=9224

TBF their profit is in decline.

That is an unnecessary interrogation, you don't need to verify the initial call at all. Simply call your bank on your own.

The standard doesn't say that you must accept all URIs up to a certain length. You're allowed to decide, based on context, what is "too long".

That is to say, you're allowed to have a double standard. And I love this sort of stuff, thanks for sharing.

I hate that I never considered this before.