Seminar
| Parent Program: | -- |
|---|---|
| Location: | SLMath: Eisenbud Auditorium, Atrium, Online/Virtual |
Mathematical AI Safety- What is it, why do it, how is it looking?
In this talk I will try to describe how I think about AI, where it's going, and why I think mathematicians can help make it safe. I will also shamelessly advertise for the Mathematical A.I. Safety Institute (MAISI) opening in January!
"AI safety" is an umbrella term referring to any kind of intervention that will enable us to co-exist safely with a powerful AI system. People think about various kinds of interventions: What kind of AI system we build and how we train it; How do we interact with our AI systems; How do we set up our world so that AI can exist safely within it.
At the moment, the kinds of theoretical tools we have for thinking about safety are highly limited. There are intuitions for what one might want in order to make a safe AI but they are nascent and mostly pre-rigorous, which makes them hard to work with and implement. That is where theoreticians - including mathematicians - come in!
This event was followed by a teatime reception from 3:00-3:30 p.m. open to all on-site attendees.
Speaker Bio: Jacob Tsimerman, Scientific Director of the Mathematical A. I. Safety Institute (MAISI), is a 2026 Fields Medalist and Professor at the University of Toronto. He has received the 2023 Ostrowski Prize, the 2022 New Horizons in Mathematics Prize, the 2019 Coxeter–James Prize, and the 2015 SASTRA Ramanujan Prize.
No Notes/Supplements Uploaded