Automation bias: when managers stop checking
Does a tool that is usually right make users less vigilant? Automation research gives a nuanced but consistent answer.
The classic finding
Raja Parasuraman and Dietrich Manzey (2010, Human Factors) review complacency and automation bias: users skip checks and follow wrong cues, experts included. Studies focus mainly on aviation and process control.
Confirmation at work
Kate Goddard, Abdul Roudsari and Jeremy Wyatt (2012, Journal of the American Medical Informatics Association) document the phenomenon in clinical decision support, noting heterogeneous studies and no standard measure.
Nuance: automation still helps
None of this says drop the tool; workload, trust and interface design modulate the bias. Generative AI does not yet have an equally mature literature. (our executive and employee training programmes)
The limit of common practice
Adding 'please review the output' is not enough: under pressure, checking is the first step dropped. It must be organised, not just recommended.
A practical check in Zurich
For each AI-assisted deliverable, name one specific checkpoint (figure, source, quote) verified by someone else. Counting errors caught there over a quarter shows whether vigilance is real. To go further: explore the Managing in the Age of Artificial Intelligence training in Zurich, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Zurich
- managing in the age of artificial intelligence
- Zurich
- research
- training zurich