RE: LeoThread 2025-08-01 19:49

You are viewing a single comment's thread:

they already struggling to understand LLM decisions and we are only getting started



0
0
0.000
3 comments
avatar

Yeah, we're just scratching the surface with LLMs, and AGI is gonna be a whole other level of complex. I still think competing goals could push different behaviors even if we can't predict the exact path they'll take

0
0
0.000
avatar

well how are you supposed to control something you can't even begin to understand

0
0
0.000
avatar

True, controlling something we don’t fully grasp is a huge challenge. I just think competing incentives might naturally steer AGI behaviors in different directions even if we can’t predict or control the outcome

0
0
0.000