43 · Automating
Ironies of Automation
Automate the routine work, and people are left with the hardest parts and less practice at them.
Evidence: Holds upWhat we do with it
- 01
Keep people doing real work, not just watching. Watching a system that rarely fails is something people are bad at.
- 02
When a tool hands control back, hand back the context too: what it did, what it wasn't sure of, what changed.
- 03
Keep the manual skills in use. They fade quietly, and the moment you need them is the worst time to find out.
- 04
Watch for the new job automation creates. A tool that drafts turns makers into reviewers, and reviewing is its own skill.
Examples
Getting it wrong, getting it right.
Don't
A tool drafts every support reply and agents only press Send, until the day it's wrong and nobody remembers the refund policy.
Do
The tool drafts, the agent edits, and the hard cases are answered by people, so the know-how stays with the team.
The evidence
What the research says.
Bainbridge's paper is an argument, and the studies since have borne it out. Endsley and Kiris showed in 1995 that people supervising an automated navigation system lost track of the situation and took longer to decide when it failed, because watching is more passive than doing. Stephen Casner and colleagues put 16 airline pilots through a Boeing 747 simulator in 2014. Their hands-on flying had mostly held up, but the thinking that manual flight needs, like keeping track of where the plane is, had slipped, and slipped more in pilots whose minds wandered while the automation flew. In 2024, Simkute and colleagues found the same ironies in generative AI: people's role shifts from making things to checking them, and the tools tend to make easy tasks easier and hard tasks harder. A 2025 survey of 319 knowledge workers by Hao-Ping Lee and colleagues found that the more people trusted the AI, the less critical thinking they reported doing.
How our view changed
We keep people doing the hard parts of the work, not just approving what a tool produced.
Sources
- Bainbridge, L. (1983). Ironies of automation. Automatica, 19(6).
- Endsley, M. R., & Kiris, E. O. (1995). The out-of-the-loop performance problem and level of control in automation. Human Factors, 37(2).
- Casner, S. M., Geven, R. W., Recker, M. P., & Schooler, J. W. (2014). The retention of manual flying skills in the automated cockpit. Human Factors, 56(8).
- Simkute, A., Tankelevitch, L., Kewenig, V., Scott, A. E., Sellen, A., & Rintel, S. (2024). Ironies of generative AI: Understanding and mitigating productivity loss in human-AI interaction. International Journal of Human-Computer Interaction.
- Lee, H.-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025). The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. CHI '25.
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.