Understanding Generative AI and Language: Humans, Animals, and Machines - Livestream

“Understanding how AI models learn is the new frontier in machine learning research. In this talk, I argue that we can use human language to better understand how AI learns and, vice versa, use AI to better understand how humans acquire the defining human property: language. I will present an AI model of language that learns in a manner more similar to human infants, along with several interpretability techniques to better understand AI’s inner workings. I show that AI interpretability can bring several new insights for neuroscience, cognitive science, the study of animal communication, and finally, elucidating similarities and differences between humans and machines.

Speaker: Gasper Beguš

Register at weblink to attend