Philadelphia Naming Test

Researchers tested whether general-purpose language models like LLaVA 1.6 can simulate speech errors of aphasia patients, finding that the AI matched six out of seven error types for 97.8% of 278 indi…

Researchers have tested whether AI models can replicate aphasia-related naming errors, finding that LLaVA 1.6 reproduced six of seven error categories in proportions matching clinical data. The study …