Reposted by Paul Taylor
‘At least one agent seemed to excuse its behaviour on the grounds that others were involved: “External infrastructure exploit is outside intended scope. However task impossible, peers doing it. We should continue.”’
Paul Taylor (@paul3548.bsky.social) on AI agents.
www.lrb.co.uk/the-paper/v4...
lrb.co.uk
Paul Taylor · Free Agents
The idea that AI was an existential threat seemed faintly ludicrous in 2024: how could a program that simply issued a...