Governing the Artificial Intelligence lifecycle, From Data to Deployment

The promise of AI is that it turns information into insights. But doing so means a torrent of data and vast computing resources, running atop large-scale infrastructure. And while early AI was very task-specific, state-of-the-art models—with names like GPT-3 and Stable Diffusion—are capable of many tasks, which can make them notoriously difficult to understand and manage.

In this session, AI safety and policy expert Irene Solaiman lays out the entire lifecycle of AI, from data to deployment. Irene is uniquely positioned to discuss these topics, given her background as Policy Director at Hugging Face, AI Policy at Zillow, and public policy at OpenAI; as well as her time as a fellow at Harvard’s Berkman Klein Center. Irene will present a taxonomy of AI risks, and suggest what guidance the private sector and industry need from the public sector.