floweronthehill

I wonder if AI can come up with mathematical conjectures. As in, they feel it's right but can't prove it. What even happened in Fermat's brain to sense it was true?

A mention of Saint Simon's memoirs in the wild is always unexpected. I'm curious to know how many people alive have read the whole thing.

These are called Fermi problems.

I've counted paper ballots for multiple presidential elections in my country.

People who think it's not safe should really spend some time learning how it works. It's impossible to cheat at scale. Each ballot is verified to be correct my multiple eyes. A person is reading, one is writing down the name, one is verifying and some other things I don't remember.

To cheat you need to have everyone in on it. A whole town involved to cheat and to at best win one polling station. It's safe because anyone can attend the counting, so each party can send someone to check no shenanigans is going on.

So the more votes you want to be winning by cheating the more people must be brought in the conspiracy. That's impossible to be unnoticed at the scale of a city, much less at the scale of a country.

I believe local llms are the future. It will only get better. Once we get to the level of even last year's state of the art I don't see any reason to use chatgpt/anthropic/other.

We don't even need one big model good at everything. Imagine loading a small model from a collection of dozens of models depending on the tasks you have in mind. There is no moat.