Emerging Algorithms and Mathematical Paradigms for Reliable Next-Generation AI

This IPAM workshop at UCLA runs from 22 to 26 February 2027 in Los Angeles and centres on the algorithms and mathematics behind reliable next-generation AI. Its particular focus is sampling, which underpins core applications in Bayesian inference and, as the organisers note, drives modern AI by powering diffusion models and large language models.

The programme gathers researchers to examine emerging algorithms and mathematical paradigms aimed at making these systems more dependable. It is an in-person event, and a request for posters is sent to registered participants, providing a route for attendees to share their own work.

The organising committee is made up of Ben Leimkuhler (University of Edinburgh), Soledad Villar (Johns Hopkins University) and Andre Wibisono (Yale University). Participation is through IPAM’s application and registration process linked on the official page, where prospective attendees should also confirm any funding or travel support. The workshop suits mathematicians, statisticians and machine-learning theorists interested in the sampling methods at the heart of contemporary generative AI.