44 · Automating
Calibrated Trust
Rely on a tool as much as it has earned, task by task. Not more, and not less.
Evidence: Holds upWhat we do with it
- 01
A tool's skill is uneven. It can be excellent at one task and wrong at the next one that looks just like it, so trust it task by task.
- 02
Show people how the tool has actually done on this kind of work. What people see it get right and wrong moves their trust as much as any stated accuracy.
- 03
Make its misses easy to see and easy to report, so trust can come down when it should.
Examples
Getting it wrong, getting it right.
Don't
An assistant that answers every question in the same confident voice, whether it's quoting the manual or guessing.
Do
One that links the source when it's quoting, and says plainly when it isn't sure.
The evidence
What the research says.
Lee and See's review found that trust steers how much people rely on automation, most of all when it's too complex to understand fully. Kevin Hoff and Masooda Bashir reviewed 127 studies in 2015 and found three layers to that trust: a person's general tendency, the situation, and what they've learned from the system itself. Ming Yin, Jennifer Wortman Vaughan and Hanna Wallach showed in 2019 that trust in a machine-learning model moved with both its stated accuracy and the accuracy people saw for themselves. The unevenness is real. In a preregistered experiment with 758 consultants at Boston Consulting Group, published in 2026, those using GPT-4 finished 12.2% more tasks, 25.1% faster and with better quality on tasks inside the AI's abilities, and were 19% less likely to get the right answer on a task just outside them.
Sources
- Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1).
- Hoff, K. A., & Bashir, M. (2015). Trust in automation: Integrating empirical evidence on factors that influence trust. Human Factors, 57(3).
- Yin, M., Wortman Vaughan, J., & Wallach, H. (2019). Understanding the effect of accuracy on trust in machine learning models. CHI '19.
- Dell'Acqua, F., McFowland, E., Mollick, E., Lifshitz, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2026). Navigating the jagged technological frontier: Field experimental evidence of the effects of artificial intelligence on knowledge worker productivity and quality. Organization Science, 37(2).
Last reviewed November 9, 2026. How we keep this current
See it differently? If you've read research that points another way, or watched this go differently in your own work, tell us at hello@synergyprod.com. We'll read it and write back.