I wonder how they build up their training set? That couldnt have been that easy.
> Facebook trained the AI by finding patterns in the words and imagery used in posts that have been manually reported for suicide risk in the past. It also looks for comments like “are you OK?” and “Do you need help?”
"Dave, you should stop thinking about killing yourself!"
- "But...I wasn't thinking about killing myself?!"
"Come on, Dave, deep down you know you want to. You can't fool me! I know everything about you."
"- Well, now that I think about it, watching Facebook all day does make me feel quite depressed..."
Curious what these were.
There was an bad issue with a place I cared enough about to try to correct, so I ventured into the "community" they setup to build their training set to try to get it fixed. It was such a depressing place. Hundreds of people tricked by gamification into working for free for a for-profit company, only to be frustrated and have their efforts shit upon by its half-ass algorithm.