Mastodon Feed: Post

Mastodon Feed

jonny@neuromatch.social ("jonny (nonvenomous)") wrote:

Meanwhile the people whose minds have been swiss cheesed by AI insist that this is because the code the LLMs write is hyper-efficient and TOO GOOD for us to read. These motherfuckers "solved" the code not working out of the box problem by strapping "a species-ending quantity of carbon emissions" worth of guess and check on top, tuned out, and when they look again they think they made a god that's too powerful to understand, rather than a burst dam of dogshit that defies all understanding with its breathtakingly relentless volume of dogshit.

[posts from @roon on twitter, 17th September] if you pause for a moment to try and read the code Astra (and I presume fable) are writing it becomes clear that “corrigibility” has become a matter of faith. they are using crazy meta-programming and abstruse primitives to write hyperefficient code we have really no choice but to ask another astra to read/use the outputs of astra 1. this is relatively recent. i think even 1-1.5 generations ago people were reading the code because there was so much that didn't work out of the box lotta people missing the point but this is me complaining that the systems are quickly becoming unmonitorable and we’re just taking them at their word