I didn't find that to be true. As a grad student and also as a postdoc, you work on whatever your PI can get grants for, which means you work on whatever the bi

I didn't find that to be true. As a grad student and also as a postdoc, you work on whatever your PI can get grants for, which means you work on whatever the big grant-funding agencies think is fashionable. Also in academia there's less ability to pivot when a project isn't working out -- if the results will be "novel" (i.e. you can get a paper out of it), you're pretty much stuck continuing in a line of work until you publish it, even if you've already concluded it won't be useful in the real world. In industry, as soon as I determine that a piece of work won't be useful, I can drop it and work on something more useful. Personally, I much prefer the criteria of "usefulness" as a reason to continue a project, rather than "novelty". Other people may differ on that preference.