Do by doing

I proposed, most recently, during a fascinating multi-faceted discussion with a Swedish AI engineer, the notion of “do by doing”. By this point, everyone has heard of learning by doing, or experiential learning—a student-centered approach where knowledge is acquired through active participation, experience, and direct engagement, rather than passive observation. I suggest that we should rather “do by doing” when it comes to realising our goals, aspirations, dreams… and how we interact with AI (specifically LLMs). To do by doing is to get the thing done, whatever it is, by doing what it takes. Don’t sit around and wonder about the best way to interact with your LLM. Don’t sit around watching YouTube videos on how to become fit, or how to cook the best meals, or how to play the guitar. No amount of video watching will teach me how to ski, ride a motorcycle, or learn a language.

My most recent personal example of this is from my Japanese studies, where I learned 2,300 kanji within six months. This is only possible by doing it. Sounds stupid, I know, but at its core, “do the thing by doing the thing” truly is the message I feel matters the most. To perfect something is to log more than 10,000 hours… doing it. Sure, you need to train to do it—by doing it.

In progress

My photo from Watches and Wonders in Geneva yesterday, where JLC really do by doing. Here’s a video of this magnificent watch.

JLC do by doing

It’s a harder pill to swallow than the more cossetted approaches of reading, watching videos, building yourself up… because it is that. It is the hard part. Doing something. Not thinking about it—doing it. Writing this post will only happen if I do it. Could it be better? Sure, probably. Would it be better if it sat in drafts for a year before I got around to making it “perfect”? No.

Do it.

Enjoyed this post? Get in touch.