Venue
Program and Schedule
Organizers
Important Dates and Announcements
Welcome

Back to Schedule

Yoshihiro Maruyama

The Australian National University, Australia, yoshihiro.maruyama@anu.edu.au

Contextuality in Artificial Intelligence and Cognitive Science

Contextuality is a transdisciplinary phenomenon that may be observed across the sciences. Contextuality in quantum physics and cognitive science has been discussed extensively in the QCQMB community, and so in this talk, we focus upon contextuality in natural language semantics, natural language processing, and artificial intelligence, which is in a stark contrast with the more traditional principle of compositionality. We explicate and articulate, inter alia, the relationships between contextuality and compositionality in natural language and artificial intelligence. We elucidate, moreover, what contextual language models in natural language processing, especially state-of-the-art Transformer models such as BERT and GPT-3, tell us about the relationships between compositionality and contextuality. We also discuss, in passing, how (contextual) linguistic cognition connects with (compositional) mathematical cognition through state-of-the-art models in natural language processing.