Exploring foundation models - Session 4

Speakers:
Professor Phil Blunsom
Head of Science, Cohere and Professor of Computer Science, University of Oxford
Professor Anthony Cohn
Turing Fellow
Katie Collins
PhD Student, Machine Learning Group
Professor Yarin Gal
Turing AI Fellow
Professor Mirella Lapata
Professor in the School of Informatics, University of Edinburgh
Professor Maria Liakata
Turing AI Fellow
Professor Percy Liang
Associate Professor of Computer Science, Stanford University
Professor Michael Wooldridge
Director of Foundational AI Research
Timestamps:
00:00 Intro
46:42 Could you elaborate more on safety metrics for your next generation of HELM?
49:35 Could you tell us more how we can navigate HELM in terms of adding new tasks and metrics?
52:54 Now you’re able to see all these metrics in one place, have you noticed any especially interesting trends?
57:08 Can you tell us more about comparing language models?
1:56:12 If you had one request for policy makers to establish one requirement or obligation for large language models, what would it be?
1:57:47 When it comes to tricky societal questions, especially in relation to bias and fairness, does the panel have any thoughts on how we, as a community, can make progress?
2:00:13 How can people from other research areas, not just in language, use these foundation models?