https://dwyer.co.za/problems-not-solutions.html


Learn about problems, not solutions

Sometimes people ask me what they should learn - they've heard AI is
kind of hot, but DevOps is too? What's the best way to get into AI?
Should they start with computer vision or NLP? Is React better than
Angular? What about Blockchain? Is it hype, a scam, or The Next Big
Thing?

While it's certainly possible to evaluate each of these areas and try
to provide some insights (AI is kind of hot, but there are a a ton of
wannabe data scientists and few open roles, while the opposite is
true for DevOps, study linear algebra and probability theory, either,
React, it's hype), I always tell people to take a step back and not
worry about any of these. They are solutions, and you shouldn't learn
to use solutions.

Managers classically ask you to "bring me solutions, not problems",
and that's all very well, but problems are definitely more
interesting than solutions. And they generally last longer too. If
you want to see what happens when you focus too much on solutions,
just take a look at all the "big data startups" and "Hadoop startups"
from several years ago - all of them were focused on a solution, and
none of them knew what problem they were solving and now they are all
gone.

I'm not saying all the AI startups are going the same way, but they
might. Especially those that are "AI startups" that don't know what
problem they are solving.

If you're starting out, your problem doesn't have to be specific. It
could still be loosely mapped to a field. For example: "I want to
help people stay healthy" -> "Medicine" -> maybe "AI advances in
medicine", but now you're grounded by something, and if AI turns out
to not be helpful, you can easily change your focus slightly and
still keep the problem that you've spent years learning about in
sight.

So pick a problem, or at least a field. If you have to get into a
solution like "AI", then make sure it is grounded by a problem. You
can pick a problem, make friends with it, and it'll keep you company
for decades or centuries. If you try to do this with a solution,
it'll likely leave you in a ditch wondering why you wasted those
years trying to get to know it.