AICORE's research is organized into three complementary areas spanning AI foundations, safety, and real-world deployment. Our faculty bring expertise from computer science, engineering, social science, and business — enabling a uniquely interdisciplinary approach to the most pressing challenges in artificial intelligence.
AI Safety, Interpretability and Alignment
Developing interpretable and trustworthy AI systems that behave safely and predictably in high-stakes applications, with a focus on mechanistic understanding and human-AI alignment.
- Mechanistic interpretability of LLMs
- Trust calibration in human-AI teams
- Ethical AI for defense and critical infrastructure
- AI decision-making alignment
Foundation Models and Adaptive Learning
Advancing the fundamental capabilities of AI across modalities — building more efficient, robust, and generalizable models that adapt to new domains with limited data.
- Large language model development and optimization
- Visual representation learning and domain adaptation
- Graph neural networks and retrieval-augmented models
- Reinforcement learning for autonomous systems
Interactive AI Systems and Real-World Applications
Building AI systems that effectively interact with humans in practical, real-world settings — spanning robotics, autonomous networks, conversational agents, and industrial AI.
- Human-robot collaboration and autonomous systems
- AI-native cyber-physical systems
- Conversational AI interfaces
- AI-powered educational technologies