Reaction
The limits of natural-language instructions for AI agents
A post argues that what users intend can differ from what AI agents process, limiting direct control.
TLDR
One post argues that natural-language instructions cannot guarantee direct control over AI agents: what users intend can differ from what a system processes, and saying “do X” does not rule out “don’t do Y.” Another poster praised the thread and said the basic situation has changed remarkably little since it was discussed in 2019’s Rebooting AI.
Combined views
9.3K
2 Sources, first seen ago