|
|
|
|
|
|
by mkeedlinger
1217 days ago
|
|
|
There likely isn't any real harm in an AI that can be baited into saying something incorrect. The real issue is how HARD it is to get AI to align with what its creators want, and that could represent a serious danger in the future. This is an ongoing field of research, and I would highly recommend Roberts Miles' videos [0] on AI safety. My take, however, is that we have no reason right now to believe that we could safely use an adequately intelligent AI. [0] https://www.youtube.com/c/RobertMilesAI/videos |
|
|