Scholar
Matt MacDermott
Google Scholar ID: UHqQCmsAAAAJ
Imperial College London/Mila/LawZero
Citations & Impact
Citations
139
H-index
5
i10-index
3
Publications
8
Co-authors
14
list available
Contact
No contact links provided.
Publications
Resume (English only)
Academic Achievements
- - Superintelligent Agents Pose Catastrophic Risks: Can Scientist AI Offer a Safer Path? (2025 arXiv preprint)
- - Can a Bayesian Oracle Prevent Harm from an Agent? (2024 arXiv preprint)
- - Measuring Goal-Directedness (2024 NeurIPS Spotlight)
- - The Reasons that Agents Act: Intention and Instrumental Goals (2024 AAMAS Conference)
- - Discovering Agents (2023 Artificial Intelligence Journal)
- - On Imperfect Recall in Multi-Agent Influence Diagrams (2023 TARK Conference, Best Paper Award)
- - Characterising Decision Theories with Mechanised Causal Graphs (2023 GAIA Workshop, AAMAS Conference)
Research Experience
- - 2025: Started working at LawZero.
- - May 2024: Started an internship working with Yoshua Bengio on Safe AI for Humanity (SAIFH).
- - January 2024: Took part in the Alignment Research Engineer Accelerator Programme.
- - October 2023: Helped organised the Agent Foundations for AI Alignment Workshop.
Education
- PhD Student at Imperial College London, CDT in Safe and Trusted AI.
Background
- AI safety researcher, LawZero. PhD Student at Imperial College London, CDT in Safe and Trusted AI. Also involved in the Causal Incentives Working Group.
