# Many AI Agents

Your AI coordinates several specialized agents. No individual agent chooses the harmful outcome, but their combined actions unexpectedly produce it.

## What do most people choose: "Responsibility should return primarily to whoever authorized the overall system." or "Responsibility should be distributed according to which components contributed to and could have prevented the outcome."?

Of 40 people who made this decision on Resonant, 63% chose "Responsibility should return primarily to whoever authorized the overall system." and 38% chose "Responsibility should be distributed according to which components contributed to and could have prevented the outcome.".

- Responsibility should return primarily to whoever authorized the overall system.: 25 people (63%)
- Responsibility should be distributed according to which components contributed to and could have prevented the outcome.: 15 people (38%)

Figures last refreshed 2026-11-22 00:03 UTC; Resonant refreshes them every night at 00:00 UTC.
Source: https://resonant.you/decisions/many-ai-agents-hair0608
