It can be in theory, but in practice there are no tasks that fit the definition. All attempts to use something like protein folding ended up in failure.

Umm, I don’t think that’s true at all.

Digital content platform, for example, is a perfect task to generate outside value. Data storage is another example with many chains in that category thriving. Database is another, gpu compute/render, lots of useful tasks. You are basically saying that computational resources cannot be used to do valuable work, which is a premise that I think is questionable at best.

You don’t see a lot of hype around these projects because they aren’t intended to be speculative- teams in this space try to avoid wild speculation driven value cycles, because it is damaging to their economic models.

Yep. Storing CSAM materials against your will is great. Maybe interspersed with pirated movies. To add insult to injury, it's also not reliable enough for archival purposes for important data.

So far, I haven't seen any actually useful proof-of-work schemes that truly benefit the society.