Recently, I read a phrase in the book Million Dollar Weekend by Noah Kagan (a book about entrepreneurship) that got me thinking. The phrase is:
Small experiments, repeated over time, are the recipe for transformation in life.
And a complementary phrase:
Experiments are supposed to fail. And should they fail, you just take what you’ve learned and try again a little bit differently. Eventual success is a byproduct of trying more things.
Those phrases got me thinking about how experimenting can add a lot of success to our learning goals in Data science or AI.
Let me explain it in more detail.
For simplicity, let’s say I want to learn a new language, and I’ll choose French because it is something I’m truly learning now.
Then, to learn French, I could take some classes, get a tutor, sign up for several language learning apps, get French-speaker friends, watch movies and series in French, read short stories in French, listen to podcasts in French, change the language of my cellphone to French or my less favorite (this one drives me crazy sometimes) change to French the app I use for my digital groceries list.
But all those choices are just possibilities, and I need to choose a few if I want to succeed.
Side note: for me, the last example is similar to what is currently happening with online AI or Data Science courses. There are thousands available, and every day, it is getting increasingly difficult to choose which ones are the right ones for us or will give us the skills we need to get a job in our desired industry. But let’s go back to our French example.
How do I know which of the previous options is best for learning French?
And more importantly, which one is the best for ME?
In short, I can only know it by trying them.
For example, I can imagine what it is like to listen to podcasts in French, but I will not be certain if that option is good for me until I try it.
For that, I can set up an experiment (sorry, the scientist in me gets very excited about experiments).
I can choose two different podcasts so I can have a variety of accents and speaking speeds. Then, I can select eight episodes (it could be more or less) and listen to a couple per week.
Then, after a month, I can tell if I’m noticing any difference in my listening capabilities or if I can recognize more words. Or, on the opposite side, I could see that I was getting more and more frustrated because I couldn’t recognize anything of what was being said, and I declared, after the eight episodes, that listening to podcasts was not the right option for me at the moment.
But for that conclusions, I had to experiment, and expose myself to the possibility that podcasts could be an excellent resource for learning a new language (or not).
And I had to try, not once, but several times, to reach reliable conclusions.
Learning from my experiment will let me recalibrate my learning journey and explore something new in my next attempt.
But all required for me to try something small consistently over time.
The other bright side of this experimenting-with-your-life scenario is that just by reframing it as an experiment, I feel less afraid of failing.
And that, my friends, gives me a huge relief.
I just have to try something new, knowing beforehand that the whole idea is to fail. Experiments are supposed to fail, but as long as I learn from my experiment, that time will not be in vain!
Eventually, I’ll find an option that works better for me, and then I can continue on that path until I reach my goal.
In a data science context, running an experiment could look like this:
- Try a new pipeline and decide if you want to switch to the new one or remain with the one you currently use.
- Trying a new library and see if that adds something valuable to our skill box.
- Adding another layer to our Python capabilities.
- Trying Computer vision for the first time and seeing what the fuzz is about.
- Take a short course about LLMs with the idea of understanding more about generative AI.
You name it; the sky is the limit!
So, what do you think of this approach? Are experiments your thing? As always, let me know if you plan to give it a try.