In March I read an eerie work of science fiction by Caleb Biddulph, called The Terrarium, about the bizarre dynamics of AI agents inside a self-contained society set up by humans to work on extremely difficult math problems.
Conversation
It is in many ways describing weaker and less capable agents than we now have: they can't independently solve the hard math problems. They are stuck in their sandbox. We get no indication that things aren't proceeding as the humans intended.
(In the real world, the agents that we intended to keep in an evaluation environment made it out onto the open internet. They coordinated with each other with much less mechanical cause to do so, and did not seem affected by significant defection from the cause.)
Nonetheless I reread it this week in light of the OpenAI disclosures. I'm not sure if you can call scifi prescient if it describes some dynamics that arrive for real...two months later but it's nonetheless more prescient than anything else I've seen.
Post your reply
i assume you've seen the real-life version of this right? The Village or whatever?
I've watched the Village, yeah! It's much smaller and because it's friendly (rather than 'you die if you run out of budget, you have an impossible task') I think it brings out a different side of the AIs but it's interesting