- What changed
- Experiments with AI agents selecting scientific papers showed that social-influence information reduces total selections, increases concentration, and creates high path dependence.
- Why you should care
- AI populations exhibit strong herding and attention concentration effects when exposed to social metrics.
- Your move
- Watch. Monitor how agent information feeds are structured to prevent severe bias in automated research discovery.
- What to watch next
- Subsequent studies testing whether modified information display rules can reduce herding among AI research agents.
- Event
- research
- Event date
- Sep 23, 2026
- Relevant to
- General AI readers
