Mind, consciousness and ChatGPT: can ChatGPT impute unobservable mental states to others?

Speaker: Paolo Tonella Abstract: In this seminar, Prof. Tonella will present the results of a recent study conducted at Stanford to assess the theory of mind (ToM) abilities of ChatGPT, i.e., the ability to impute unobservable mental states to others. The authors administered classical false-belief tasks, widely used to test ToM in humans, to ChatGPT. Based on their findings, the authors conjecture that a ToM-like ability might have emerged spontaneously as a byproduct of ChatGPT’s language skills.