Now that the AI overlords demand more and more training data, maybe we will start "paying" by annotating some unit of information, aka Captcha-style.
back
1 comments
You’re describing google.
Though training data sets based on private information about people are worth more to them, so they focus on that.
That lets them do things like run crooked ad auctions that screw websites and advertisers, intentionally worsen search results exactly enough to maximize profits, and other stuff that came out during their trial.
I understand your point. But in my hypothetical model, the paying entity (the customer) does not have an incentive in manipulating the search results, and the search is still optimised (only) for relevance to the user query.