Twitter’s content moderation AI bots keeps confusing rockets with the male genitalia- Technology News, Firstpost

Mehul Reuben DasJan 06, 2023 11:52:58 ISTAfter firing most of the staff in addition to contractors who moderated content on Twitter, Elon Musk had his staff rely closely on AI bots to do the job. As a results of this, a number of regular photographs and tweets bought taken down or had their attain restricted, unintentionally. One hilarious consequence of Twitter utilizing AI bots to average content, nonetheless, has solely not too long ago come to the floor.
Twitter’s content moderation AI bots have been coded in such a approach that they will take down a publish and even droop an account if they’re 95 per cent sure {that a} publish has damaged the platform’s guidelines.
Speaking to Quartz, a former Twitter worker who left the firm not too long ago, revealed that the AI bots that Twitter is utilizing, keeps confusing photographs of rockets with that of the male genitalia. Not solely that, accounts that posted photographs of rockets have been suspended from the platform.
For instance, Spaceflight Now, and person Michael Baylor, each of whom stay stream launches for NASA, have been all locked out of their accounts after posting content a few SpaceX launch on Tuesday. Similarly, Starbase Watcher, which tracks exercise at SpaceX’s Texas facility was additionally locked out.
Similarly, Spaceflight photographer John Kraus stated his account was additionally suspended after he shared a video of NASA’s Artemis I launch. All 4 accounts have since been unlocked.
In response to a tweet by Baylor’s colleague about his account being suspended, Musk replied “looks as if our picture recognition wants some work!”
Quartz interviewed an nameless ex-employee who used to work on Twitter’s content moderation methods. They instructed the outlet the instruments had been recognized to misidentify acceptable photos for pornographic content. This may embody, for instance, a pedicure photograph that comprises plenty of flesh-colored pixels.
The AI bots have been coded in such a approach that they will take down a publish and even droop an account, if they’re 95 per cent sure {that a} publish has damaged the platform’s guidelines,
Musk and his staff cracking down on porn on the platform is undoubtedly an amazing factor. But as is the case with all the pieces else that Muks has executed with Twitter since he took over, the execution has been a catastrophe. 
If Twitter actually desires to rely much less on human moderators, it ought to scale back the precision thresholds of the AI bots and machine studying instruments that it makes use of to differentiate delicate content, in order that it doesn’t flag each publish as inappropriate.

https://news.google.com/__i/rss/rd/articles/CBMijgFodHRwczovL3d3dy5maXJzdHBvc3QuY29tL3RlY2gvbmV3cy1hbmFseXNpcy90d2l0dGVycy1jb250ZW50LW1vZGVyYXRpb24tYWktYm90cy1rZWVwcy1jb25mdXNpbmctcm9ja2V0cy13aXRoLXRoZS1tYWxlLWdlbml0YWxpYS0xMTkzOTQ0Mi5odG1s0gGSAWh0dHBzOi8vd3d3LmZpcnN0cG9zdC5jb20vdGVjaC9uZXdzLWFuYWx5c2lzL3R3aXR0ZXJzLWNvbnRlbnQtbW9kZXJhdGlvbi1haS1ib3RzLWtlZXBzLWNvbmZ1c2luZy1yb2NrZXRzLXdpdGgtdGhlLW1hbGUtZ2VuaXRhbGlhLTExOTM5NDQyLmh0bWwvYW1w?oc=5