AI and learning: why 'friction' is a feature, not a bug

Yes, it seems so, even if everyone today is looking for the easiest path. AI is excellent at giving us immediate answers, but the process of solving a problem ourselves is crucial for memory and deep understanding.

Who would've thought? A bit of healthy struggle, that moment your brain fries trying to find a solution, isn't a bug. It's a "feature" of the system, as Yechiel K. argued on dev.to on May 26, 2024. This is especially true for those studying code or complex subjects.

It depends on how you use it: AI can be a great crutch if you just want the answer, but a poor tool if you want to understand. Many simply ask the bot for solutions, missing the opportunity to develop critical thinking.

Instead of having AI write the code or report, we could ask it to explain why a certain approach works, or help us find errors. Many tech education experts today recommend using AI to explain code, not just generate it. Not quite the same thing, is it?

The mentor is no longer just the person who gives you answers, but the one who teaches you to interact well with AI. They show you how to formulate the right questions and, most importantly, how to doubt the answers you receive.

Imagine a running coach: they don't run for you, but give you tips on technique. Similarly, the mentoring approach in the age of AI is shifting towards teaching "prompt engineering" and critical verification. We need to learn to think with AI, not let it think for us.