Trustworthy Multimodal Interaction
AI capable of learning causal models of its environment and interacting with people to communicate and update those models.
AI systems don’t just read text. They interpret images, speech, movement, and context — often all at once. This area of study focuses on helping AI make sense of that mix of information in ways that are accurate, safe, and aligned with human needs.
We explore how AI can learn from experience, adjust when new information appears, and clearly explain what it understands — and what it doesn’t. The goal is to create systems people can trust and work with confidently across healthcare, education, research, and everyday life.
Learning from Many Forms of Input
Bringing together language, images, sound, and other signals to build a fuller understanding of what’s happening. AI that draws from multiple sources can better interpret complex real-world situations.
Explaining What It Knows
Designing AI that can clearly describe its reasoning and acknowledge uncertainty. When systems communicate openly, people can make informed decisions and stay in control.
Adapting Over Time
Ensuring AI can update its understanding as new data or feedback becomes available — without becoming unpredictable. Responsible learning over time is essential for real-world use.
