Design / prototype
The Third Chair: Rotating multi-model crisis evaluation
Two models negotiate from conflicting positions while a third mediates. The roles rotate so the study can separate assigned position from model behavior.
Research program
Research
Currently studying local AI-assisted cybersecurity and what happens when AI systems face conflict, incomplete evidence, adversarial pressure, and consequential choices.
Questions in scope
Current lines of inquiry
This program grows out of my work in cybersecurity research, curriculum, and applied technical education.
Adversarial model behavior
Evaluation of escalation, deception, bargaining failure, and brittle judgment across changing roles and incentives.
AI-assisted security testing
Using models to find weak assumptions, overlooked paths, and failure modes in conventional cyber systems.
Digital trust
Studying when uncertainty, explanation, and abuse resistance justify reliance on an AI-supported decision.