Brandon Mossop
CAUSAL AI · BAYESIAN INFERENCE · AGENT-BASED MODELLING · WORLD MODELS
Building AI systems that learn and reason about causation with robustness and uncertainty quantification.
I work on causal structure learning, causal forecasting, Bayesian uncertainty, and agent-based simulation. My research asks how AI systems can move beyond prediction toward models that explain mechanisms, support interventions, and remain useful when environments change.
Research areas
Causal Discovery
Learning causal structure from observational data, statistical tests, and contextual knowledge.
Causal Forecasting
Using causal structure to improve forecasting robustness, interpretability, and intervention reasoning.
Causal Simulation & Agents
Combining causal models, Bayesian inference, and agent-based simulation for policy and complex-system analysis.
Selected work
Regime-aware causal Bayesian forecasting
A forecasting framework for non-stationary time series that combines regime-specific temporal causal discovery with Bayesian prediction and uncertainty quantification.
Agentic causal discovery
Multi-agent causal discovery systems that combine statistical evidence with language-model reasoning and critique.
Learning behavior for agent-based models
An emerging research direction exploring whether causal discovery and probabilistic structural functions can learn behavioral rules for agents directly from data.
Current questions
My current work is centered on a few practical questions:
- Can causal structure make simulation models more trustworthy under interventions?
- How should uncertainty propagate from learned causal mechanisms to system-level policy outcomes?
- Can agents learn useful causal world models through interaction rather than passive observation alone?
- How can small language models contribute context without replacing statistical evidence?