A Skill With No Practice

A skill holds as long as you use it. A tool that does the work for you also takes away the load the skill stood on.

Doctors and colonoscopies

In 2025 a medical journal published an observation of 19 experienced doctors, each with more than 2,000 procedures behind them.

The comparison was between procedures the doctors performed without help from software: 795 before that help was brought into daily work and 648 after.

The share of procedures where the doctor found a dangerous growth fell from 28.4 to 22.4 percent. Same doctors, same equipment; what changed was the habit of working with a prompt on screen.

The authors add a caveat: the study is observational and does not prove a cause. The direction, though, is visible.

Students and essays

In another study, 54 students spent four months writing essays three ways: with a chat service, with a search engine, and with no tools. Their brain activity was recorded with instruments.

In the group using the chat service, connections in the regions responsible for memory and attention were weaker. Eighty-three percent of that group could not quote the essay they had just handed in.

The most useful finding was about order: those who wrote on their own first and brought the model in afterward did better than those who started with the model.

This is about students and academic essays, not about your work. The numbers do not transfer directly, but the order of operations is worth remembering.

Trust and verification

A survey of 319 knowledge workers showed a pattern that concerns all of us.

The more a person trusts the tool, the less they verify. The more confident they are in their own abilities, the more verification they do.

The work itself changes as well: instead of solving the task, a person checks someone else's answer, stitches pieces together, and supervises the process. That is a different skill, and it is the one that now needs practice.