It has never worked though, as far more people would vote helpful if something made them laugh (because people just think “more upvote”)
I dunno, that’s why they added the “Funny” button and I see people use that all the time. Even the nearly-useless “Was this review helpful?” section on Amazon has some use to a customer making a purchasing decision.
If ML can be used to further help the issue, what’s the problem? At least “AI” is being used for something that’s actually trying to solve a practical issue in an attempt to improve the platform and not as an immediate way to extract maximum profit with minimum effort.
You could argue that Valve loves to automate its customer service to save money, and that would be valid and true, but I think improving the platform experience by trying to reduce (if not eliminate) unhelpful reviews is good.
