Feb 20, 2026 Feb 20, 2026 2026-02-20

Today, a social network (I won’t name names) reminded me that 7 years ago I completed Stanford University course on cryptography. It wasn’t an easy course, I still remember the long nights trying to understand some pretty hard calculus problems, but, all in all, I thoroughly enjoyed it.

This wasn’t the only course I took from Coursera. For the next couple of years I did a deep dive in Artificial Intelligence, or Machine Learning, as it was called back then. I started from the wonderful Andrew Ng foundations module, and I went all the way up to the fascinating Generative Adversarial Networks one. I won’t put all of them here, if you’re curious you can have a peek at my LinkedIn certifications page.

Back then, artificial intelligence wasn’t spectacular. There was no ChatGPT yet, and you would just try to solve difficult problems in Jupyter notebooks, and get yourself accustomed to Matlab, or python and matrix multiplications. You would learn complicated neural networks architectures and try to find the gradient descent of a cost function. Then you would apply Adam optimizers and do regression tests.

If all of this sounds like Chinese to you, it’s on purpose. Just stay with me a little bit more.

A couple of years later, a benign chat app was released by a non-profit startup called OpenAI, under the name of ChatGPT. The world literally changed overnight. ChatGPT exploded, and now it has a staggering 800 million users every month. See, a relatively small advancement in machine learning, called transformers, made all the Chinese above instantly usable. Practical. Easy to understand.

Where am I going with this?

Well, when I started to learn artificial intelligence, 7 years ago, the field was still obscure. There was little to no practical mass adoption. People interested in this had to make significant cognitive efforts to understand what’s going on. But when the thing reached critical mass, those who started early were incredibly well positioned.

Why? Because they understood the foundations. They understood how this “magical” chat starts from optimizing a cost function to match features with outcomes. They knew how it was all optimized. And they could start performing on this field, at a very high level, instantly.

I openly admit to be a power AI user. I enjoy vibe coding iOS apps in coffee shops and I made my own always-on agent, AIGernon. I applied my cognitive framework, Assess Decide Do, on top of Claude (and a handful of other LLMs).

All this while still living location independent, and being blessed with a one year old child.

If you’re still here, I’ll infer you want to know more about how this worked for me, so here’s a short excerpt of my experience with continuous learning.

How To Get The Best of Continuous Learning

This doesn’t apply only to artificial intelligence. It can be applied to language learning, creativity, or any other thing that, stacked on top of the previous one learned, will eventually make you a polymath – an being a polymath is surprisingly beneficial.

Start early, start when it’s hard

If you want to be well positioned, start early. Start when the field is not yet mainstream. Start when it’s difficult. Learn the basics, and try to build on top of them, even though the results may not be spectacular first. I cannot stress this enough: the thing you’re learning will never unfold in the way you expected it to be. It will unfold in a much better way.

For example, when I started to learn Korean, I expected to get a language certification. Instead, I won a hackathon in South Korea.

Build lateral skills

By “lateral skills” I mean skills that can complement some of your already established expertise. Deep specialization is commoditized these days – AI is already an always-on source of deep knowledge. What’s missing, though, and what AI cannot easily replace, is the peculiar mix between unusual skill combinations. Pair this with real life experience, and you’ll be unstoppable.

For example, I am a coder for 35 years, but artificial intelligence is so much more than coding. So, knowing how to code definitely helps me, and, compared with your weekend vibe coder, it puts me in a much better position.

Make it a gravitational habit

Don’t try to make it a separate event. Don’t put it on your New Year’s resolutions. Don’t make vision boards with it. Instead, make continuous learning as boring as brushing your teeth every day. I call these kind of habits “gravitational habits”, because the more you do them, the more you will do them, regardless of how you feel.

Gravitation still pulls you no matter how you feel. Happy, motivated, bored, tired, a gravitational habit will always pull you towards it. If you want to know how I apply this to finances, I wrote a tiny guide called Gravitational Habits for Financial Resilience, give it a try.

Experiment, experiment, experiment

This is not a fixed term task. The name itself say it very clearly: continuous learning. That means, beyond making it a part of your lifestyle, that some of the stuff you learn will be eventually replaced by something new – and there’s nothing you can do to stop this. So, what worked yesterday may or may not work today. This reality forces an experimental mind, one that thinks in scenarios, not in frozen paths.

Keep experimenting and try to detach yourself from the result. The goal of an experiment is to learn something, not to reach a specific goal. If you do reach a goal, though, be happy, you earned it.

Where To Go From Here?

Well, literally wherever you want. Continuous learning will open new paths, many of them unthinkable with your current level of understanding.

Trust the process.