MATS Program - LessWrong
blogCredibility Rating
Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.
Rating inherited from publication venue: LessWrong
MATS is one of the primary structured entry points for new researchers entering the AI safety field, with alumni contributing across interpretability, agent foundations, and alignment research areas.
Metadata
Summary
The MATS Program is an independent AI safety research and education initiative that mentors emerging researchers through workshops, talks, and connections to the SF Bay Area and London AI safety communities. It aggregates outputs from multiple cohorts, including notable work on mechanistic interpretability, sparse autoencoders, and alignment theory. MATS serves as a key pipeline for developing the next generation of alignment researchers.
Key Points
- •Provides mentorship, workshops, and talks to emerging AI safety researchers across multiple annual cohorts
- •Connects participants with established AI safety research communities in SF Bay Area and London
- •Has produced notable research outputs including SolidGoldMagikarp, sparse autoencoders, and mechanistic interpretability work
- •Serves as a talent pipeline and field-building initiative for the broader AI alignment research ecosystem
- •Aggregates 300+ tagged posts representing research from MATS program participants
Cited by 1 page
| Page | Type | Quality |
|---|---|---|
| MATS ML Alignment Theory Scholars program | Organization | 60.0 |
Cached Content Preview
x This website requires javascript to properly function. Consider activating javascript to get access to all site functionality. MATS Program — LessWrong MATS Program Edited by Ryan Kidd , et al. last updated 12th Mar 2026 The Machine Alignment, Transparency, and Security (MATS) Program is an independent research and educational seminar program that provides emerging researchers with mentorship, talks, and workshops and connects them with the SF Bay Area and London AI safety research communities. Add Posts Subscribe Discussion 3 Subscribe Discussion 3 Posts tagged MATS Program Most Relevant 8 675 SolidGoldMagikarp (plus, prompt generation) Ω Jessica Rumbelow , mwatkins 3y Ω 208 8 72 SERI MATS Program - Winter 2022 Cohort Ω Ryan Kidd , Victor Warlop , Christian Smith 3y Ω 12 7 334 Understanding and controlling a maze-solving policy network Ω TurnTrout , peligrietzer , Ulisse Mini , Monte M , David Udell 3y Ω 28 6 21 Project proposal: Testing the IBP definition of agent Jeremy Gillen , Thomas Larsen , JamesH 4y 4 5 123 Soft optimization makes the value target bigger Ω Jeremy Gillen 3y Ω 20 5 71 SERI MATS - Summer 2023 Cohort Aris , Ryan Kidd , Christian Smith 3y 25 5 69 SERI ML Alignment Theory Scholars Program 2022 Ω Ryan Kidd , Victor Warlop , ozhang 4y Ω 6 5 63 How MATS addresses “mass movement building” concerns Ryan Kidd 3y 9 4 186 Finite Factored Sets in Pictures Ω Magdalena Wache 3y Ω 35 4 120 Taking the parameters which seem to matter and rotating them until they don't Garrett Baker 4y 48 4 26 Talk: AI safety fieldbuilding at MATS Ryan Kidd 2y 2 3 141 Recontextualization Mitigates Specification Gaming Without Modifying the Specification Ω ariana_azarbal , Victor Gillioz , TurnTrout , cloud 5mo Ω 15 3 118 Efficient Dictionary Learning with Switch Sparse Autoencoders Anish Mudide 2y 20 3 112 I found >800 orthogonal "write code" steering vectors Jacob G-W , TurnTrout 2y 20 3 105 Predictions for shard theory mechanistic interpretability results Ω TurnTrout , Ulisse Mini , peligrietzer 3y Ω 10 Load More (15/302) Add Posts
5b6927a2a7c55d21 | Stable ID: NmM4NDMzMG