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.
Autonomy in academia depends entirely on the field, the funding situation, and the advisor. If you choose wisely, you can build a highly autonomous career. But it is hard for the inexperienced to figure out how to make the right choices.
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.