The mastery study
Why it matters Getting the task done and learning how to do it are different goals. AI help can serve the first while quietly costing you the second.
In a randomised study, developers learned to use a Python library that was new to them, some with an AI assistant and some without (Anthropic, Shen and Tamkin 2026). Afterwards everyone took a test on what they had learned. The group with AI help scored about 17 points lower.
The average hides the interesting part. Not everyone who used the AI lost out. The people who used it to understand (asking why something works, asking for explanations, checking their own grasp) kept much more of what they learned than the people who handed it the work.
Delegating: asking for the result. "Write a function that does X." You get code, and the thinking happens somewhere else.
Engaging: asking in a way that makes you think. "Why does this use a lock here?", "What would happen if two calls arrive at once?", "Here is how I think it works: what did I get wrong?"
What this module does
You will practise engaged ways of asking, predict what AI-written code does before you run it, and then solve two twins of earlier problems with the tutor switched off. The twins test whether the idea stuck, not whether you remember the code.
Recap AI help cost mastery on average, but not for people who used it to understand. How you ask matters more than whether you ask.
Spotted a problem in this lesson? Report it · Ask about this lesson
This exercise opens soon
Finish the lesson before it and this one unlocks. The whole course is free, no account needed, and you can read this lesson right now.
This browser is blocking cookies for learn.floati.life, so lessons you finished here cannot open in order. Allow cookies for this site and reload.