AIThe Decoder1 day ago
AI search agents fail at asking clarifying questions
1 min read
AI search agents fail at multi-step research not from poor searching but from not asking clarifying questions on ambiguous queries. DiscoBench shows models that search repeatedly instead of asking follow-ups perform worse at 51.9 percent than those that guess. Removing ambiguity boosts accuracy by up to 40 points.
Level
Hype check
Tap to vote and see what everyone thinks.
