In Academia (at-least in CS where I did a PhD), you have to be working on a novel idea. It doesn't matter how many people actually care about the problem you are working on and/or will be impacted - as long as the idea is new and you are the first one.
Completely opposite in the startup. The problem has to be relevant. It doesn't matter if 5 other startups before you have worked on a similar problem - as long as you can out execute them.
This switch requires some serious "unlearning"