Better evaluations for more capable agents

Boston Working Group

A small recurring research group for people actively working on AI agent evaluation.

Date
September 30, 2026 · 6–8 PM
Location
MIT · Cambridge, MA
Participation
Application required · Limited to 10 participants

About

A recurring, in-person research group

This is the first meeting of a recurring Boston-area group for researchers actively building or studying AI agent evaluations.

The immediate goal is to help participants get to know one another and build a shared agenda for future meetings.

First meeting

Meet the group and shape what comes next

01

Get to know one another

Share backgrounds, current work, and the evaluation questions each person is thinking about.

02

Build the agenda together

Identify the problems, research directions, and collaborations worth carrying into future meetings.

Who should apply

For people doing the work

Applicants should have hands-on AI agent evaluation experience, including evaluations, agent systems, or related empirical research.

The group is limited to 10 participants. If this meeting does not work for you, or you want to hear about later sessions, send us a general expression of interest.