#6 - The 3 Steps to Learning & Why You Can Now Skip Step 3.

AI’s changing how we must teach ourselves and future generations new skills.

Follow me on this…

There are three steps to learning a new skill.

1. Learning what it is.

2. Learning why it matters.

3. Learning how to do it.

When I got my first engineering job, I learned how to use Kubernetes — a framework that makes it easy to deploy and maintain micro-service applications. My learning process followed the three-step process above.

1. Learning what it is — I heard of Kubernetes for the first time shortly after starting work, since my team was already using it

2. Learning why it matters — My colleagues made it clear that it was important for my job and future career that I know how to work with Kubernetes.

3. Learning how to do it — I read a textbook on Kubernetes.

(1) always comes before (2) — I couldn’t have understood why Kubernetes mattered if I didn’t know it existed in the first place.

And (2) always comes before (3) — If I didn’t think Kubernetes mattered, I wouldn’t have wasted a minute of my time learning how to use it.

Most educational content focuses on step 3.

Online courses and textbooks make the valid assumption that (1) you’ve already heard about the tool/technology they’re teaching you about, and (2) you think it's important enough to learn how to use, which is why you picked up the course/textbook in the first place.

Still they’ll usually open with a section about the what and the why, just to get you hyped up. After that though, the remaining 90% of the content is focused on step 3 of the learning process: teaching you how to actually do the thing.

My Kubernetes textbook, for example, began with a chapter on the technology’s history and why the software world needs Kubernetes. But the rest of the book divulged the nitty-gritty implementation details.

This teaching approach makes sense. It pays to be an expert. If you know the C++ standard library inside and out, you’re going to be compensated very well for your services.

But this might be changing…

AI is shifting the value from step 3 to steps 1 & 2.

I write mostly Typescript and Python code during my day-to-day. I rarely write C++ and on the odd occasion that I do, I write it slowly and poorly.

But nowadays, I no longer need to have mastery of C++ to write pretty good C++ code. Github Copilot can correct my syntax and handle the nitty-gritty optimizations.

In a future where AI agents take over most of the rote implementation duties, our value as workers will be in our ability to make sound decisions; to exercise impeccable judgement of what is worth working on and which approaches provide the most acceptable trade-offs for our desired outcomes.

We must learn how to form opinions, understand tradeoffs spar with our ideas, and make principled decisions.

And what that means is that we need to take the “T” shaped approach to skill acquisition — develop as wide an understanding of our domain as possible and dive deep on a small number of topic-areas when needed.

Put simply, we need to focus a lot more on steps 1 and 2 in our learning process.