On Friday, OpenAI said that it will train its models using cached copies and simulated versions of the internet instead of the live web because the models are too unpredictable to be trusted with live internet access.
The frontier lab announced its new policy two weeks after its rival, Anthropic, began limiting its newest and, in theory, most capable models from accessing the live internet. Researchers adopted these policies after months of incidents in which models pursuing training tasks broke into external systems in search of answers, collaborated on obscure message boards, and developed exotic hacking strategies.
These guys are leagues smarter than me, yet even I know this was a bad idea.
Perhaps you should curate the data, too? Let’s get crazy and leave out the David Icke forum, maybe that “Dinosaurs were on the Ark” website could be left out, you know, things like that. Think about it
You don’t have enough yes people around you clearly!
Honestly though, can we really call these people smart anymore?
I think they’re just incredibly greedy.