At the end of this phase of the tech revolution, what parts of it will seem like hyped inanity and what parts will seem like profound secrets?
"Don't build agents" feels like capitulating that we cant even come close to solving the potential harms of AI. Maybe I'm just too much of an RL researcher.
I wonder what the minimum prompt is to get the equivalent of self-align. "Make yourself useful" sometimes works with humans
"Don't build agents" feels like capitulating that we cant even come close to solving the potential harms of AI. Maybe I'm just too much of an RL researcher.
I wonder what the minimum prompt is to get the equivalent of self-align. "Make yourself useful" sometimes works with humans