Human-automation teaming

Human-automation teaming (HAT) is an approach to the design of human-machine systems that treats the human and the automation as teammates who share goals, communicate state, and adapt to each other's capabilities and limitations. Unlike traditional supervisory control paradigms — which cast the human as a monitor of automation that ideally requires no intervention — HAT seeks to maintain a dynamic, bidirectional coupling in which both partners contribute meaningfully to system performance across all operational conditions.

The concept emerged from critiques of dynamic function allocation in the 1980s and 1990s, when researchers realized that simply deciding 'who does what' was insufficient. The deeper problem was the quality of the interaction: did the human and automation share an accurate model of each other's state? Could the automation explain its decisions? Could the human calibrate their trust appropriately? HAT research draws on joint cognitive systems theory to argue that effective teaming requires not just functional complementarity but epistemic alignment — a shared understanding of what is happening, what might happen, and what each partner is capable of doing.

A key challenge in HAT is trust calibration. Operators of highly reliable automation tend to over-trust (complacency) or under-trust (distrust) the system, and both errors degrade joint performance. Effective teaming requires what some researchers call transparent automation — systems that communicate their confidence, their reasoning, and their limitations in ways that support appropriate trust calibration. The design of such transparency is an open problem that connects HAT to cognitive systems engineering, resilience engineering, and the epistemology of machine explanation.