TY  - RPRT
TI  - Language-model groups overstate consensus when replaying human deliberation on a reasoning task
AU  - Tengfei Shao
PY  - 2026
UR  - https://arxiv.org/abs/2609.20543
ID  - 2609.20543
ER  -