Algorithm Aversion
Edited by Paweł Raja, PhD · Published
Definition
Algorithm Aversion is the phenomenon where individuals lose trust in algorithmic decision-makers faster than human decision-makers after seeing them make a mistake, even when the algorithm performs better overall.
- Also known as:
- Algorithmic aversion, Distrust of algorithms, AI aversion
- Category:
- Decision Architecture
- Impact:
- High
- Landmark study:
- Dietvorst et al. (2015)
- Evidence here:
- 2 field findings, 6 cited studies
Why it happens: the psychology of algorithm aversion
Algorithm aversion is the tendency to abandon an algorithm after seeing it make a mistake, even when it is more accurate overall than the human alternative. Dietvorst, Simmons and Massey (2015) showed that people who watched a forecasting model err became much less likely to use it — more so than after seeing a human make similar errors.
One explanation is that people expect algorithms to be perfect and treat any error as proof the system is broken, whereas human errors feel forgivable and correctable. People also believe humans can learn from mistakes and handle unusual cases.
Aversion is stronger for tasks seen as subjective or uniquely human. Castelo, Bos and Lehmann (2019) found people resist algorithms for tasks such as judging a joke or a date more than for analytical tasks, and Longoni, Bonezzi and Morewedge (2019) found patients resist medical AI partly because they feel their case is unique.
Control reduces aversion. In a follow-up study, Dietvorst, Simmons and Massey (2018) found people were far more willing to use an imperfect algorithm when they could adjust its forecasts, even slightly.
Algorithm Aversion examples
A forecaster chooses to manually adjust or entirely discard statistical models after seeing them predict a minor stock drift incorrectly, reverting to less accurate gut-based decisions.
- A clinician may discount a diagnostic model after one visible error, even when it outperforms them on average.
- A driver overrides a navigation app that once routed them poorly, trusting their own memory instead.
How to avoid algorithm aversion
- Compare error rates, not single errors. Ask how often the algorithm and the human are wrong over many cases, rather than reacting to one visible mistake.
- Keep a human in the loop where it adds value — for example, allowing small adjustments — instead of choosing between full automation and none.
- Hold humans and algorithms to the same standard. If you would forgive an expert for an occasional miss, apply the same tolerance to a model with a better record.
How to design for it (nudge strategy)
Restore trust by letting users make small adjustments to the algorithm's output (a limited 'control override'). Dietvorst, Simmons and Massey (2018) found that even slight control made people far more willing to use an imperfect algorithm.
Ethical use: design for choices people would endorse on reflection — a nudge, not sludge. Be transparent and keep opting out easy.
The evidence (2)
- AI Assistance in Radiology: Why AI Predictions Did Not Improve Radiologists' Accuracy
Healthcare · Average deviation from the diagnostic standard (0-100 probability points; lower is better): X-ray only (control mean) 22.6 → With AI predictions 22.9 (+0.3 pts; Randomized experiment; 227 radiologists; 324 patient cases; 41,920 observations)
- Overcoming Algorithm Aversion: Letting People Slightly Modify an Algorithm's Forecasts
Technology · Participants choosing to use the algorithm's forecasts (Study 1): Algorithm could not be changed 32% → Could adjust each forecast by up to 10 percentiles 76% (+44.0 pts; Lab experiment; Three incentivized experiments: Study 1 in the Wharton Behavioral Lab; Studies 2-3 online (Study 2: 816 participants))
Key studies
Seeing an algorithm err made people choose a less accurate human forecaster instead.
Dietvorst, B. J., Simmons, J. P., & Massey, C. (2015). Algorithm aversion: People erroneously avoid algorithms after seeing them err. Journal of Experimental Psychology: General, 144(1), 114–126. DOI →
Letting people make small adjustments to an algorithm's output greatly increased its use.
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. DOI →
Aversion was stronger for tasks perceived as subjective and weaker when algorithms were described as human-like.
Castelo, N., Bos, M. W., & Lehmann, D. R. (2019). Task-dependent algorithm aversion. Journal of Marketing Research, 56(5), 809–825. DOI →
Patients preferred human providers to equally accurate AI, driven by concern that AI neglects their unique characteristics.
Longoni, C., Bonezzi, A., & Morewedge, C. K. (2019). Resistance to medical artificial intelligence. Journal of Consumer Research, 46(4), 629–650. DOI →
Key research
- Algorithm Aversion: People Erroneously Avoid Algorithms After Seeing Them ErrBerkeley J. Dietvorst, Joseph P. Simmons, Cade Massey · Journal of Experimental Psychology: General (2015)
- Human Trust in Artificial Intelligence: Review of Empirical ResearchElla Glikson, Anita Williams Woolley · Academy of Management Annals (2020)
- Algorithm Appreciation: People Prefer Algorithmic to Human JudgmentJennifer M. Logg, Julia A. Minson, Don A. Moore · Organizational Behavior and Human Decision Processes (2019)
- Combining Human Expertise with Artificial Intelligence: Experimental Evidence from RadiologyNikhil Agarwal, Alex Moehring, Pranav Rajpurkar, Tobias Salz · NBER Working Paper Series (2023)
Related biases
Algorithm Aversion: frequently asked questions
What is algorithm aversion?
Algorithm aversion is the tendency to stop trusting an algorithm after seeing it make an error, and to prefer human judgment even when the algorithm is more accurate on average.
What causes algorithm aversion?
High expectations of perfection, the belief that humans learn from mistakes and handle exceptions better, a desire for control, and the sense that one's own case is unique.
How can algorithm aversion be reduced?
Giving people some control over the output, showing comparative track records, and explaining how the algorithm works all reduce aversion.
What is the opposite of algorithm aversion?
Algorithm appreciation — the tendency, documented by Logg, Minson and Moore (2019), to give more weight to advice labeled as coming from an algorithm than from a person.
Cite this page
Raja, P. (Ed.). (2026, 8 Oct 2026). Algorithm Aversion. Behavioral Economics Lab. https://www.behavioraleconomicslab.com/biases/algorithm-aversion