4 ms·
> But WhatsApp moderators told ProPublica that the app’s artificial intelligence program sends moderators an inordinate number of harmless posts, like children
by RcouF1uZ4gsC 5y ago
> But WhatsApp moderators told ProPublica that the app’s artificial intelligence program sends moderators an inordinate number of harmless posts, like children in bathtubs. Once the flagged content reaches them, ProPublica reports that moderators can see the last five messages in a thread.
It seems like whatever the AI decides is questionable gets sent to moderators automatically, also revealing the thread history (at least partially) as well.
- spullara 5y agoI think the AI was applied against non-E2E encrypted data like groups.
- bilal4hmed 5y agothis says that when you report a message, it sends over the last 5 messages https://twitter.com/WABetaInfo/status/1435221936888483847 https://twitter.com/WABetaInfo/status/1435221936888483847 The pro publica article is pure click bait
- barbazoo 5y agoThis doesn't at all imply that the same isn't happening upon automatic detection of "reportable" messages.
- KaiserPro 5y agoIts all about signal to noise. THose who are claiming that they are running Machine learning at the edge, think about this: how on earth can they trust the data coming from it? We all know that ML is hard, we also know that to get accurate classification requires serious horsepower. None of this can be run on phone hardware without people noticing. If you were to design a system for handling reported data, who's hard constraint is _human_ eyeball time, would you run ML at the edge? No. You'd make it so the "report this message" sent the last n lines and the attachment, and then run the ML optimised for accuracy (not speed or memory) on your hardware that you control and trust explicitly. I would have thought all of this is obvious, given that facebook is all about avoiding spending money on human moderators. Edge ML is not going to help you do that.