Hm, can you explain what you mean? My initial reaction is that AI oversight doesn’t actually look a ton like this position of the interior where defenders must defend every conceivable attack whereas attackers need only find one successful strategy. A large chunk of why I think these are disanalogous is that getting caught is actually pretty bad for AIs — see here.
See also the position of the interior.
Hm, can you explain what you mean? My initial reaction is that AI oversight doesn’t actually look a ton like this position of the interior where defenders must defend every conceivable attack whereas attackers need only find one successful strategy. A large chunk of why I think these are disanalogous is that getting caught is actually pretty bad for AIs — see here.