Rare, Relevant, and Still Wrong: How Memory Fails LLM Group Decisions
The paper studies which information should survive when an LLM group's discussion is compressed into a short memory for a final decision maker. It introduces CrossProfile, a diagnostic benchmark of 60 templates (480 tasks) that crosses fact ownership while holding text fixed. Across Qwen2.5-7B, Llama-3.1-8B, and two DeepSeek models, selection by relevance readmits decoys, rarest-first selection is no better than random on average, and the decision model gives extra weight to facts attributed to many agents. Removing attribution tags raised accuracy in the classic configuration. Full transcripts yielded lower accuracy than no notes in natural discussions.
Open original source huggingface.coOne primary source; other links are context, not independent corroboration.
- Primary sourceHugging Face Daily Papers
- Discovery linkHugging Face
- Linked in sourceView arXiv page
- Linked in sourceView PDF
The paper page is linked. Peer review and findings have not been independently checked by AINewsIQ.
Open full article view →- Original paper+45
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
The paper studies which information should survive when an LLM group's discussion is compressed into a short memory for a final decision maker. It introduces CrossProfile, a diagnostic benchmark of 60 templates (480 tasks) that crosses fact ownership while holding text fixed. Across Qwen2.5-7B, Llama-3.1-8B, and two DeepSeek models, selection by relevance readmits decoys, rarest-first selection is no better than random on average, and the decision model gives extra weight to facts attributed to many agents. Removing attribution tags raised accuracy in the classic configuration. Full transcripts yielded lower accuracy than no notes in natural discussions.
A READER'S EVIDENCE CHECKHow to read this signalSix questions about claims, sources, risks, and open checks
This is a signal-level guide until claim-level evidence is available. “Not recorded” means the evidence has not been assessed here; it does not mean a result is false. How we check AI news →
- 01Who is making each claim?
The claimant and whether each claim was measured or asserted are not recorded in this public brief.
- 02What was actually tested?
Not recorded per claim. Check the source for sample size, tasks, model and benchmark versions, and reproducibility.
- 03Where are the sources?
Original link: huggingface.co. Brief basis: fetched source page. Primary page date and quote locations are not established in this panel.
- 04What could mislead?
Results are reported for four specific models and a diagnostic benchmark of 60 templates; source claims not independently checked.
- 05What is at risk if wrong?
Not assessed per claim. Consider safety, privacy, cost, and adoption decisions before relying on the result.
- 06What would change the view?
No claim-specific check is recorded. An inspectable primary method or independent replication could change the assessment.
Key findings from the stored brief
- Relevance-based selection readmits decoys because decoys are on topic, losing 23 to 31 points to rarest-first in the classic configuration.
- Rarest-first selection is no better than random selection on average and wins only where rare facts are the useful ones.
- Removing or randomly reassigning attribution tags raised accuracy by 13 to 24 points in the classic configuration.
- In natural discussions, the complete transcript yielded lower accuracy than no notes at all.
Results are reported for four specific models and a diagnostic benchmark of 60 templates; source claims not independently checked.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
The paper page is linked. Peer review and findings have not been independently checked by AINewsIQ.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
A publication revision history is not stored for this signal. New engagement snapshots do not mean the source itself changed.
Why this may matter
Multi-agent systems often compress group discussions into memory before a final decision. The findings suggest that common selection heuristics and attribution signals can mislead the decision model, so memory design needs to consider how evidence is balanced, represented, and attributed.
Limits and open questions
Results are reported for four specific models and a diagnostic benchmark of 60 templates; source claims not independently checked.
Organization-specific financial, security, and operational effects require the organization's own context and review.
Discussion
Comments are reviewed before publication. Discuss the evidence, not the person.
No published comments yet.
Reader discussion opens with accounts. You can read the original source and published comments now.