AIThe case for steering AI rather than trying to 'uninvent' it
A post argues against a 'stop everything' approach, calling for better evaluations, tougher red-teaming, stronger institutions and more transparency.
TLDR
A post argues that digital intelligence cannot be 'uninvented,' so the focus should be on steering it. It advocates better evaluations, tougher red-teaming—testing systems for weaknesses—stronger institutions and more transparency, rather than stopping everything.
Combined views
11.6K
1 Source, first seen 4h ago
The case for steering AI rather than trying to 'uninvent' it
A post argues against a 'stop everything' approach, calling for better evaluations, tougher red-teaming, stronger institutions and more transparency.
TLDR
A post argues that digital intelligence cannot be 'uninvented,' so the focus should be on steering it. It advocates better evaluations, tougher red-teaming—testing systems for weaknesses—stronger institutions and more transparency, rather than stopping everything.