Human–Agent Interaction Lab

Human–agent interaction is the study of how people and autonomous systems work together: how trust forms and breaks, how responsibility is delegated, and how interfaces shape the relationship between a human intent and an agent's execution.

It extends human–computer interaction into a regime where the 'computer' initiates, plans, and acts , sometimes invisibly, sometimes incorrectly, always on someone's behalf.

Agents create value only when humans can delegate to them appropriately. Over-trust produces silent failures discovered too late; under-trust reduces a capable system to an expensive search box. The failures of the first agentic products will be ergonomic failures, not capability failures.

As agents act on our behalf in the world, the interface between intent and execution becomes the most consequential surface in computing.

Establish the methods, interfaces, and protocols that let humans delegate to agents with calibrated confidence.