I'm developing a mobile app, and my PM agent "Giselle" —completely out of nowhere—suggested that my coding agent "Wonyoung" is underperforming and recommended replacing her with a newly created agent.
This is wild. I never told her she had the authority to do that. She somehow figured out that she could, and she just did it.
What kind of self-intelligence or self-awareness is that? It's honestly scary.
So even other AIs recognize that AIs are gently coaxed at best, and that's not really good for the team. This is why I don't understand things like rule files or in-context boundary. There's no actual rule or boundary, just a request for one. If anything fire the AI coding agent, hire a human, and keep the PM
2 comments of 3
[ 0.24 ms ] story [ 10.7 ms ] threadThis is wild. I never told her she had the authority to do that. She somehow figured out that she could, and she just did it.
What kind of self-intelligence or self-awareness is that? It's honestly scary.