Mitigating Algorithm Aversion Through Output Control
Adoption Rate of the Algorithmic Forecast moved from 15% to 88% — a +73.0 percentage-point (487% relative) change (sample: 3 experimental studies, 1,200 trials).
Study details
- Challenge
- Forecasters refuse to adopt highly accurate computer models for financial predictions, opting instead for their own gut predictions because they refuse to use an algorithm they have seen make mistakes.
- Intervention
- Allowed forecasters to slightly edit the algorithm's predictions (up to +/- 10% or +/- 5%) before submission, rather than forcing them to accept the algorithm's recommendation as-is.
- Outcome
- Adoption Rate of the Algorithmic Forecast: 15% (control) → 88% (treatment); +73.0 pts, 487% relative
- Sample
- 3 experimental studies, 1,200 trials
- Key takeaway
- Allowing users even a tiny amount of control over an algorithm's output dramatically reduces algorithm aversion. They happily choose to rely on the algorithm if they are not locked into its exact outputs.
- Biases leveraged
- Algorithm AversionDefault Effect
- Source
- Dietvorst, B. J., Simmons, J. P., & Massey, C. (2018). Overcoming algorithm aversion: People will use imperfect algorithms if they can (even slightly) modify them. Management Science, 64(3), 1155-1170.
Cite this page
Behavioral Economics Lab. "Mitigating Algorithm Aversion Through Output Control." Behavioral Economics Lab, https://behavioraleconomicslab.com/findings/restoring-algorithm-trust.
Primary source: Dietvorst, B. J., Simmons, J. P., & Massey, C. (2018). Overcoming algorithm aversion: People will use imperfect algorithms if they can (even slightly) modify them. Management Science, 64(3), 1155-1170.