close menu
2.S988: Games, Learning, and Security
Class Information
Artificial IntelligenceInternationalNavalSystems
Graduate | 12 Units| Prereq:
Lecture:
TR 9:30AM - 11:00AM - 12:30PM (4-163) []
Modern decision-making systems increasingly operate in contested environments, where performance depends not only on managing uncertainty and resource constraints, but also on the actions of other intelligent and adaptive agents. Critical infrastructure, autonomous sys-tems, defense applications, and algorithmic markets must anticipate competition, withstand attacks, adapt to changing conditions, and continue functioning under disruption. Designing for these settings requires understanding not only how individual decisions are optimized, but also how strategic behaviors emerge when humans and artificial agents interact.
This course develops the modeling and computational foundations for designing multi-agent systems in contested environments. Students will learn how to model strategic interactions, analyze competition and cooperation among agents, understand how learning agents adapt over time, identify vulnerabilities created by adversarial behavior, and design systems that remain effective under uncertainty and disruption.
The course integrates four perspectives:
• Strategic modeling: how agents reason, compete, cooperate, and influence one another.
• Learning and adaptation: how AI and human agents update their strategies through experience and interaction.
• Security and adversarial reasoning: how systems are attacked, exploited, and de-fended.
• Resilient design: how systems can anticipate, withstand, and recover from failures and strategic surprise.
Concepts from game theory, optimization, learning theory, and network science are intro-duced as needed and applied throughout the course. The central organizing element is a semester-long computational wargame, in which student teams design a multi-agent sys-tem, implement a simulation environment, introduce strategic and learning agents, evaluate attack and defense strategies, and redesign the system for improved performance. Systems are evaluated not only by modeling sophistication, but by their ability to achieve their ob-jectives under competition, adaptation, and disruption.
2N6 students will be given additional examples with emphasis to Naval contexts.
Instructor Information
Themistoklis Sapsis (Fall)
William I. Koch Professor; Professor of Mechanical Engineering; Director, Center for Ocean Engineering; Director, MIT Maritime Consortium
Interest Areas
- Quantification and prediction of extremes events in complex systems
- Reduced order modeling of chaotic and turbulent systems
- Fluid dynamics in engineering and geophysics