It’s a study done by METR, they constantly pump out papers talking about the “significant catastrophic risks via AI self improvement or rogue replication”
The fact that they’re publishing something negative is what’s interesting here. Except that they’re reporting on a 6 month old study lol
It’s a study done by METR, they constantly pump out papers talking about the “significant catastrophic risks via AI self improvement or rogue replication”
The fact that they’re publishing something negative is what’s interesting here. Except that they’re reporting on a 6 month old study lol