Trust, but Verify: AI Safety from the Lab to the Real World

​Trust & Safety Panel — Wednesday, August 26 @ 6:15 PM

​📍 Location: 525 Market St, San Francisco, CA 94105

​The MIT AI Conference Research Slam is a fast-paced showcase of exciting AI research from across academia and industry. Researchers present their work in short, accessible talks focused on the problem, the core idea, the evidence, and why it matters.

​This year’s Research Slam is centered on Trust & Safety, highlighting work on how we evaluate, secure, govern, and responsibly deploy increasingly capable AI systems. Trust, but Verify is part of the lead-up to the Slam, bringing together experts to explore the questions and tensions shaping the field before researchers take the stage.

​In preparation for the Research Slam, this panel will help set the stage by surfacing the key challenges, perspectives, and open questions in Trust & Safety that will frame the research presented at the Slam. Attendees will get valuable context ahead of the showcase and a deeper understanding of the issues researchers are working to address.

​Our panelists will be Prof. Irene Chen (UCSF, MIT ’22) and Dr. Abubakar Abid (Hugging Face, MIT ’15, ’16). Together, we’ll explore Trust & Safety from different angles—from model evaluation and alignment to bias, high-stakes applications, and the challenges that emerge when AI systems are deployed in the real world.

​Agenda

• ​6:15 PM — Doors open; pizza, drinks, and informal networking

• ​6:45 PM — audience prompt/poll

• ​7:00 PM — Moderated panel conversation

• ​7:40 PM — Audience Q&A

• ​8:00–8:30 PM — Informal conversation and networking

​🍕 Pizza and drinks will be served.

​Arrival Instructions: Please check in with the building security desk located in the lobby of 525 Market Street. After checking in, the lobby team will direct you to the elevators and provide access to the 32nd floor, where a member of the Goodwin team will greet you upon arrival.