AI Center Impact Talk by Vanessia Wu - "Balancing Innovation and Responsibility: An Insider's Perspective on AI Safety"

The EPFL AI Center is pleased to invite you its first Impact Talk, by Vanessia Wu, Former Senior Product Manager on the Trust and Safety team at YouTube, where she was in charge of Generative AI safety, Comment/Livechat safety, Hate Speech and more.

The talk will be followed by a light lunch.

Title
Balancing Innovation and Responsibility: An Insider's Perspective on AI Safety

Abstract
When we think about AI safety, our minds often gravitate toward challenges like misinformation and deep fakes. One area that is less discussed but top of mind in large social media platforms such as YouTube/Google is: “who is ultimately responsible”? For example, say a creator used genAI tools provided by one of these platforms and published gender-biased content. Who is accountable: the creator, the platform, or the LLM model developer?
These questions matter greatly from business, legal and regulatory perspectives, which in turn affect how safe these organizations design their genAI products to be. Drawing on her extensive experience leading AI responsibility initiatives at YouTube, the speaker will shed light on how large tech companies try to balance LLM innovations with responsibility considerations. 

Bio
Vanessia was a Senior Product Manager on the Trust and Safety team at YouTube, where she was in charge of Generative AI safety, Comment/Livechat safety, Hate Speech and more.
She crafted YouTube’s GenAI Responsibility principles, led the team that pioneered LLM-based detection, and advised YouTube’s first GenAI product launches including DreamScreen (video) and DreamTrack (music). She spent 10 years at Google in the Bay Area, including with Google Search, Android and Africa.
Prior to that she worked in strategy consulting and startups. She recently relocated to Switzerland and now works for Daydream, an AI-powered search and discovery engine for e-commerce.