The One-Word Census: Answer-Choice Conformity Across 44 Language Models
Factual evidence
What the source reports
Research tested 44 LLMs on open-ended single-word prompts, finding significant answer convergence (e.g., "serendipity" 41% of the time).
Inspect the evidence
- Inclusion basis
- Enterprise AI
- Publisher and source type
- arXiv cs.CL — Computation and Language · RESEARCH
- Published by source
- 9 October 2026
- Collected by OneBench
- 15 Jul 2026, 21:08 UK
Stored source excerpt
arXiv:2607.12796v1 Announce Type: new Abstract: When a language model must pick one answer from a large space of equally valid options, which does it pick…
Short excerpt from the collected text, not the full source. Use the source link to read it in context.
The factual summary is a OneBench synthesis, not a quotation or independent verification. Collection time is not publication time. Open the source for its full context; related reporting can share the same underlying announcement.
OneBench interpretation
Institutional assessment
So what
This study on LLM answer convergence highlights an emerging model risk: a lack of diversity in outputs that could lead to systemic blind spots or unexpected correlations in downstream applications.
Do what
Your model risk team needs to account for this type of emergent behavioral bias when evaluating LLMs for critical decision support or content generation, especially in scenarios requiring diverse outputs.