My take: never solve a general problem. There are no general problems.

Individual humans have specific problems. If you spend enough time talking to individual humans, you will start to see patterns. Those generalizations are things you make up in your head, a convenient way for your brain to handle and incredibly complex world. They are false, but a pretty useful kind of false. They help you think and talk about the world, and can help you generate testable hypotheses.

So what should you do? Go talk to people about their problems, their annoyances, their frustrations. Just listen and ask good questions, without a lot of judgment. Eventually they will mention a problem where your developer-brain will say, "Oh, maybe I could help with that."

Which people should you talk to? Pick people you will enjoy serving, and pick a domain where you can make use of your strengths. I knew a guy who was working on software for printing shops to manage their workflow. His dad owned a print shop, and as a kid he worked in it. Right now I'm doing the user research for an app for medical residents; a friend runs a medical residency program and talked about a big problem they had. A possible solution fits with an interest I've had for years.

Once you've found the people and the problem, solve it for very specific people and a very specific problem. If you can knock it out of the park for them, only then look at how you can generalize the solution for other people, other problems. Google, for example, started out as an excellent search engine for stanford.edu pages, and then betaed as an excellent search engine for Linux-related web sites. Facebook started as a solution for one college, the one the founder was attending. Only when it worked there did it broaden to other colleges. And only once it dominated colleges did it expand to the broader population.