Matt MacDermott

I'm currently an Astra Fellow on the technical AI safety team at Coefficient Giving.

Before that, I was a research scientist at LawZero and a PhD candidate at Imperial College London (I'm currently awaiting my viva). During my PhD I worked with the Causal Incentives Working Group.

Papers

Reasoning Under Pressure: How do Training Incentives Influence Chain-of-Thought Monitorability?
Matt MacDermott*, Qiyao Wei*, Rada Djoneva*, Francis Rhys Ward*

arXiv preprint, 2025
Superintelligent Agents Pose Catastrophic Risks: Can Scientist AI Offer a Safer Path?
Yoshua Bengio, Michael Cohen, ... Matt MacDermott, ... et al.

arXiv preprint, 2025
Can a Bayesian Oracle Prevent Harm from an Agent?
Yoshua Bengio, Michael K. Cohen, Nikolay Malkin, Matt MacDermott, Damiano Fornasiere, Pietro Greiner, Younesse Kaddar

UAI 2025
Measuring Goal-Directedness
Matt MacDermott, James Fox, Francesco Belardinelli, Tom Everitt (2024)

NeurIPS 2024

Spotlight Paper
The Reasons that Agents Act: Intention and Instrumental Goals
Francis Rhys Ward, Matt MacDermott, Francesco Belardinelli, Francesca Toni, Tom Everitt

AAMAS 2024
Discovering Agents
Zachary Kenton, Ramana Kumar, Sebastian Farquhar, Jonathan Richens, Matt MacDermott, Tom Everitt

Artificial Intelligence, 2023
On Imperfect Recall in Multi-Agent Influence Diagrams
James Fox, Matt MacDermott, Lewis Hammond, Paul Harrenstein, Alessandro Abate, Michael Wooldridge

TARK 2023

Best Paper Award
Characterising Decision Theories with Mechanised Causal Graphs
Matt MacDermott, Tom Everitt, Francesco Belardinelli

Games, Agents and Incentives Workshop; AAMAS 2023