3 ms·
To be fair > Importantly, this is just one of many (autonomous) sessions where Claude worked on discovering new ideas. Many sessions resulted in no new discove
by porridgeraisin 3mo ago
To be fair
> Importantly, this is just one of many (autonomous) sessions where Claude worked on discovering new ideas. Many sessions resulted in no new discoveries; other follow-up sessions improved on the insight developed in this one. This document was produced by having Claude rewrite the chain of thought to include more detail to make it easier to read.
- tedbradley 3mo agoGood point. So basically, they likely spent 1+ million USD letting Mythos run for hours upon hours, and in this one case, a query with sizable compute used, apparently US$100k at the end of it all, came up with something interesting. And that $1 million was just an example. What if they ran 40 such queries? That'd be $4 million bucks flushed down the toilet. Not exactly an amount people will pay for a 1/40 chance to see if an LLM can innovate. They'd rather pay teams of humans, say, $4 million in total, since they have a much better track record for the money spent.
- porridgeraisin 3mo agoYes, see https://news.ycombinator.com/item?id=49161078 https://news.ycombinator.com/item?id=49161078