Definition
Once you have used AI on a task, you reach for it more often on the tasks that follow, and your read of how much it actually helped gets less accurate at the same time. Two things move together: the habit grows, and the judgement that would check the habit weakens. In Yu's experiments, people who had just used AI went on to use it on 44.5 percent of later tasks, against 27.7 percent for people who had worked unaided.
What is at stake
It builds inside one working session rather than over months. An afternoon is enough to leave you reaching for AI more often and reading its value less accurately.
Where the word came from
- Automation biasMosier and Skitka, 1990s
- The crossover pointParasuraman and Manzey, 2010
- The session-level loopYu et al., 2026
How episcope came to it
Yu's efficiency-gain illusion experiments, 2026. The same three pre-registered experiments that produced the efficiency-gain illusion, 2,691 people.
Papers and articles behind it
Yu, S., Jurafsky, D., Hawkins, R. D., et al. (2026). The efficiency-gain illusion: People underestimate the rate of AI use and overestimate its benefits on simple tasks. arXiv:2605.22687.