Models & Research
Two studies find language models process speech and absorb bias much like humans
A Nature Communications paper links LLM layers to the brain's timeline for understanding speech, while an arXiv study finds models show human-style in-group bias.
