About
I’m interested in next-gen agentic AI systems that incorporate robust world models into their architectures.
World models quantify an agent’s uncertainty in its beliefs, provides transparency for us to view how
an agent models its world, and in turn allows for safer operations and better value convergence.
This is one of the technologies foundational to our architecture at the GAIA Lab.
Our mission is to unify systems through federated active inference,
converging on ever-improving world models that are crafted with LLM-in-the-loop processes.
Ultimately, this compiled encyclopedic “Github of world models” bolsters a symbolic world representation
for improved and safer AI operations.
I also write about various AI subjects and interests on my blog Quarry.
Overall, I would say I’m a sober moderate regarding AI threat models.
My concern bends towards nearer-term harms (dis-/misinformation, malicious actors,
socioeconomic disparities and power concentration, etc.) vs. catastrophic x-risk scenarios
(though I don’t rate all x-risk events as sufficiently ignorable).
I’m currently undertaking an industry-PhD with the nuclear experimental lab Framatome GmbH in Germany and
researching with Prof. Dr. Thomas Kopinski at the South Westphalia University of Applied Sciences.
We are engineering AI tools for the nuclear decommissioning process as part of Germany’s energy transition.
Previous to this, I received my MSc. Physics from the University of Waterloo under Jeff Chen, investigating
machine learning methods on confined liquid crystal systems.
When I can, I enjoy taking my Olympus OM-1 along with me outdoors for some urban and nature photography.
Works & Publications
- M. Walters, R. Kaufmann, F. Neubürger, T. Kopinski, D. Marković, The CRISTAL Method: Fast, reliable analytical problem-solving with pre-synthesized grounded world models (2025). Read pdf. (submitted to NeSy 2025)
- M. Walters, R. Kaufmann, J. Sefas, T. Kopinski. Free Energy Risk Metrics for Systemically Safe AI: Gatekeeping Multi-agent Study (2025). Read pdf.
- RiskRay (Patented). Using ML to detect hips at risk of fracture.
- Traffic prediction Graph Neural Net. View the proof-of-concept toy model.
- My MSc. thesis: Machine learning topological defects of liquid crystals in two dimensions. Available online on UWSpace (2019).
- M. Walters, Q. Wei, J. Z. Y. Chen. Machine learning topological defects of confined liquid crystals in two dimensions. (Phys. Rev. E 99, 062701 / Read PDF) (2019).
- M. B. Bennett et al. Detailed study of the decay \(^{31}Cl(\beta\gamma)^{31}S\). Physical Review C 97 (6), 065803 (2018).
- E. Aboud et al. Toward complete spectroscopy using \(\beta\) decay: The example of \(^{32}Cl(\beta\gamma)^{32}S\). Physical Review C 98 (2), 024309 (2018).
- [Editors’ suggestion] M. B. Bennett et al. Isospin mixing reveals \(^{30}P(p,\gamma)^{31}S\) resonance influencing nova nucleosynthesis. Phys. Rev. Lett. 116, 102502, (2016).
- M. B. Bennett et al. Isobaric multiplet mass equation in the \(A = 31, T = 3/2\) quartets. Physical Review C 93 (6), 064310 (2016).
- C. Wrede et al. \(\beta\) Decay as a Probe of Explosive Nucleosynthesis in Classical Novae. Physics Procedia 66, 532-536 (2015).
