TY  - RPRT
TI  - Extremely coarse learning objectives induce human-aligned representations in AI vision models
AU  - Yash Mehta
AU  - Michael F. Bonner
PY  - 2026
UR  - https://arxiv.org/abs/2605.05556
ID  - 2605.05556
ER  -