Vague Queries Crash Deep Research Agents; More Search Calls Don't Help, Paper Finds
yuz9yuz · x · 2026-11-03
A new paper (arXiv:2610.39413) introduces UNDERSPEC-DR, a framework for evaluating how underspecified user queries affect deep research agents (DRAs):
- People tend to ask AI more casually than humans. Even when just one detail is missing from a query, DRA answers can substantially fall short of users' information needs.
- Notably, answers remain grounded in cited evidence — they look reliable but miss the user's intent.
- Analysis of intermediate search behavior explains why: underspecification pushes DRAs toward broader searches that lean less on the original query and more on learned priors, which may misalign with user intent.
- Takeaway: making more search calls barely helps; what DRAs need is the ability to proactively ask for clarification — recognizing when additional user input is required.
More from coding & agent
- Agent-built data pipeline flags law firms whose clients later hit SEC/DOJ trouble — doodlestein · 2026-11-03
- MIT-licensed frontend skill gives coding agents a 7-step design and verification workflow — Extra-Highlight9183 · 2026-11-03
- Trajectory launches post-training platform with GS1 and Continual Graders research previews — simonguozirui · 2026-11-03
- SMBs are rebuilding their entire GTM stack with agent-native tools, says new thread — garrytan · 2026-11-03
- Compacting Claude Code at 400K instead of 967K cut this dev's cost per request by 43% — Awkward-Let-4628 · 2026-11-03
- Capy 0.7.0 ships Captains orchestration GA; YC's Garry Tan reports 5X issue-fixing throughput — garrytan · 2026-11-03