Facebook's AI Approved Ads Claimed To Be Paid For By Terrorist Group ISIS

Although there has been a lot of praise for Facebook's artificial intelligence in its ability to discriminate against fake news, one that has the most obvious fake placements is ignored and undetected. Again, the biggest weakness remains in Facebook's AI.

A test was conducted by Vice News and Business Insider that Vice News was able to post political ads for Mike Pence (VP USA), ISIS (a terrorist organization) and Tom Perry (DNC Chairman). Meanwhile, Business Insider was able to post an ad for Cambridge Analytica, which has now been dismantled. But advertising for Hilary Clinton was blocked and not posted. Facebook claims that they do not allow any political ads from foreign advertisers. Nevertheless, VICE News managed to post an identical advertisement advertised by Russian employees in 2016.

The ads that Facebook's AI was supposed to rule out

We see the vulnerability of Facebook's AI, things that are not explicitly programmed for is likely to pass the rules of it. A properly programmed and guaranteed AI must be able to block the ads for Tom Perry and Tom Perry as active politicians without having to be told. Hilary Clinton's advertising was blocked just because it had been programmed to do so. Surprisingly, Facebook programmers seem to have forgotten the world's most dangerous terrorist organization – ISIS.

Remarkably, this is not so sophisticated terms of AI as image recognition, rocket science, speech or video processing. These are just the basic findings based on the text and the script but Facebook's AI performance was very poor.

Another fake ads, this time it was paid by ISIS - a very dangerous terrorist group

Facebook has always been at the bottom of the AI development race in the past two years by RFM. It shows that Facebook's AI does not grow much better than any other digital ecosystems. Facebook can't get past stage 1 when it comes to creating and deploying AI to improve their digital life services. Which means it is better to hire a human army to analyze its data and improve its AI as well because it is now extremely outdated and stupid.

Moreover, this work cannot be handled solely by humans, because the negative and inappropriate content should be seen and determined immediately when it is posted. While most Facebook employees have worked hard to flag the allegedly or fake content, it often tends to become trended before being erased. Economically, when human capital is really expensive, (as reducing EBIT margins indicate) they can not do it as well as the machines are rigidly programmed.

And this is the reason why the biggest program of Facebook by far is still AI, and while Facebook is trying to improve it, it seems still not enough.