AI will not save Facebook moderators from PTSD

Even if they repeat it loud and clear, artificial intelligence does not perform the miracles promised on the networks, on the contrary, since it makes mistakes and deletes, for example, YouTube channels that did nothing wrong. Of course it is not useless and facilitates the daily life of moderators, but until AI becomes the magic solution, human moderators who must do a (difficult) part of the job. They see horrors all day long, whether it is scenes of violence, pedophilia, rape and other graphic content.

Recently, more and more of these moderators have decided to express themselves and explain their difficulties in their profession. The consequences on their mental and even physical health are sometimes dramatic, and some even go so far as to commit suicide. Where is the promised miracle solution?

According to James Grimmelmann, a law professor who spoke to the Washington Post, “AI won’t solve Facebook’s problems but it will solve Zuckerberg’s” by passing responsibility to someone else. Worse still, the AI may never be able to do what is promised. Professor Goldman questions, in particular, the ability of filters to detect undesirable content (hate speech, racism, homophobia etc.) in the meshes of the net and the researcher Caplan does not necessarily see the future in a positive light.

“AI will improve in terms of contextual understanding, but there is no indication that AI will soon be able to do it better than a human being,” and they do not anticipate this change for several years. “There is no easy solution, we will continue to see problems.” This is not very encouraging, but it is nevertheless necessary to highlight the progress made in this area, but until humans know how to solve the problem reliably, the machine will not be able to do so either…

In any case, these more or less self-proclaimed giants of communication allow horrible people to express themselves with impunity and employ people who lose their minds doing the moderation, waiting for the miracle of the AI to happen, which, if it happens, will take several more years. Will things be even more out of control by then?