Sep 8, 2018, 09:42am
Why We Still Need Human Moderators In An AI-Powered World
Kalev Leetaru
As global criticism of Facebook and Twitter’s content moderation efforts reached a crescendo over the last few years, the companies have responded by going all-in with AI-powered moderation, arguing that the only way to actually properly review all of the content posted to their platforms is to use algorithmic filtering. Unfortunately, as the companies have quickly learned, machine learning tools are extremely powerful at recognizing objects and thematic portrayals in images and text, but struggle to understand the all-important context in which they are portrayed, which differentiates an iconic war photograph that defined a conflict from a criminal sexual image that should be removed immediately. These moderation struggles were back in the news the past two weeks in two cases demonstrating both the need for increased algorithmic prefiltering, but also the false mythology of all-knowing AI put forth by the companies and the very real need to continue to have humans in the loop.
https://www.forbes.com/sites/kalevleetaru/2018/09/08/why-we-still-need-human-moderators-in-an-ai-powered-world/#5ecbf4d61412
