Glossary · Term

MATS

← all terms

Definition

Plain language

A training program that helps new researchers get into AI safety work.

As stated in the literature

ML Alignment & Theory Scholars — a research fellowship program; several papers in this corpus list MATS-affiliated authors collaborating with frontier labs.

Why it matters: It helps bring new talent into AI safety research, widening the pool of people working on making AI systems trustworthy.

For example, a recent graduate interested in AI safety might join this program to work alongside experienced researchers and frontier labs.

Heard on the show

“Six authors, mostly from the alignment world — Oxford, Toronto, TruthfulAI, MATS, a couple of Anthropic and Berkeley connections.”
Episode 043 — When 'This Is False' Doesn't Stick: Why Models Learn the Lie Anyway

Mentioned in 2 episodes

  1. 043
    When 'This Is False' Doesn't Stick: Why Models Learn the Lie Anyway
  2. 007
    Exploration Hacking: When Models Sabotage Their Own RL Training