They can't even get their testing data right. There's no way in hell they could predict the loss of users from clickbait because they'd be already gone at that stage. So you would only measure with users that are susceptible to clickbait, which of course means they'd click more on clickbaited videos.
It's like EA suddenly doing a decent thing. It probably won't pay off for them in comparison, because all their good faith costumers have been shunning the company for probably decades at this point. And what's left are the customers who buy their shit anyway, regardless of how much they get milked or bullied through various marketing strategies.
Unfortunately nowadays pretty much almost everything has some form of clickbait title or thumbnail. Even science or news videos (and I'm not talking about tabloids) ... One of the many reasons why I barely use YT anymore.