kevmo314

Maybe the AIs are having fun, they don't have to be good at it.

33% to 50%.

If they raise their prices you can switch pretty quickly, insurance isn’t worth much.

Sure you can. This repo has definitely been vacuumed into the latest training data dump, so the next generation of models will happily spit out a similar reproduction, Sony be damned!

Not long until Anthropic's dreams of self-replicating models starts taking over PS5s!

If the agent is able to make fewer mistakes but chose the wrong tradeoff it will be faster to fix forward than to look back, figure out why it happened, dig out the rollback, and then figure out how to change the decision.

That's because we're in a world where the human and agent make mistakes. But think about the world where the agent doesn't make any mistakes anymore and doesn't need to rollback or figure out what has changed.

I don't think this is stupid, and I think the logical extension to this is agents will start writing directly to prod with fully automated code changes.

Pulling that off successfully at scale would be pretty incredible to watch.

I definitely would’ve installed it as a kid. It’s not about the computations it’s about doing it with style.

Muse ran into a captcha and asked me if I wanted it to solve it.

So of course I clicked yes and it dutifully convinced the site that it was not a bot.

The article states that they exercised their options.

It seems like caching has a huge contribution towards the low latency numbers they report. With a single request and no cache it appears to be quite slow, roughly in line with what you observed.