Why AI sucks, example 2,738,402
Clearly Tumblr has spent money on some kind of machine learning tool to scan drafts and label them as potentially not, umm, See Spot Run level. Yeah, I get the fact that the lawyers get paid to keep the company out of trouble and are biased towards overprotecting vs. (what to non-lawyers might seem) a more balanced approach, but the audience here are not kindergarteners, and the parameters have long seemed to be set by blushing Victorian schoomarms for whom the sight of a bit of ankle would be a scandal.
MY POINT: If you're going to spend the money on this, why on earth didn't you make it so that it tells you what part of the post the parameters triggered on? What am I supposed to do, edit it one bit at a time until the trigger stops happening? How long do I have to wait to find out if it's acceptable to the algorithmic morality cop?
I know there's, like, five people left working here so thank you for your service and maybe this can be added to the no-doubt-gigantic pile of enhancement requests.
