Gašper Beguš

Gašper Beguš is an Associate Professor in the Department of Linguistics at the University of California, Berkeley, where he integrates linguistics, artificial intelligence, and cognitive science. He leads the Berkeley Speech & Computation Lab and serves as the Linguistics Lead for Project CETI, focusing on decoding sperm whale communication. He is also the College Principal of Bowles Hall, UC Berkeley’s oldest residential college.

Research Interests: His work spans computational linguistics, phonetics, phonology, deep learning, and cross-species communication. He develops interpretable AI models—such as 'artificial babies'—that learn speech from raw audio, enabling comparisons between human and machine cognition. His research explores the origins of language, AI interpretability, and the legal and societal implications of understanding animal communication.

Publication Trends: His recent publications focus on using generative adversarial networks and convolutional neural networks to model phonological and phonetic learning, interpret neural network layers, and decode animal communication. Themes include unsupervised learning, speech encoding, rule emergence, and cross-modal comparisons between artificial and biological systems.

• Scientific Awards: (No specific awards listed in source materials)

Advising and Grants: While specific students and grants are not listed, Beguš leads a major research lab and a high-profile interdisciplinary project (Project CETI), indicating active mentorship and substantial grant funding. He collaborates with institutions and researchers globally and engages in public outreach through media and invited talks.

Labs and Teams: He directs the Berkeley Speech & Computation Lab, which develops models of language learning and AI interpretability. He is a key leader in Project CETI, an international team using AI to decode sperm whale codas. He also collaborates with InterpretAI to translate AI transparency into industry applications.