Calibrated trust
Definition
Calibrated trust is trust that is appropriately related to evidence about a system’s capabilities and limitations in a particular task.
Explanation
The question is what gives us reason to rely on an answer. A confident tone is weak evidence on its own. The problem, the system’s known limitations, and checks of the actual response can give us a more useful basis for judgment.
Trust and reliance are related but distinct. A person may trust a tool and still check its work because the consequences matter. Appropriate checking also depends on the person’s knowledge and the task, rather than a single reliability score.
In my work, capability evidence is an instructional motivation: it may help students ask where checking is especially important. That is a design argument, not a demonstrated effect of my instruction.
Scope and evidence
Calibrated trust is an established construct in human-AI interaction. A recent review is background, not its origin. No study on this site establishes that a subject-level benchmark profile changes students’ trust.