The Failure of AI - Facebook's Zuccing

“The amount of violations of human rights in a country is always an inverse function of the amount of complaints about human rights violations heard from there. The greater the number of complaints being aired, the better protected are human rights in that country.” - Daniel Patrick Moynihan (attr)

Keywords: paradox

***

Recently I've been zucced twice so fast that I am quite sure it was based on an algorithm.

The first time, I posted "Nah. Men are evil" and got zucced in under 20 minutes:



Interestingly, when I posted some followups like "Men are evil /s" or with some other indicators of sarcasm, those comments didn't get zucced.

On Sunday, I made the mistake of sharing an article from the Huffington Post, Women Are Evil.

In under an hour (within minutes of the post going out - exactly how many I don't know, but it was very quick) I was zucced:



Given the speed at which both zuccings happened and their absurdity (the first time I was being sarcastic [which hopefully would've been clear given my other comments in the thread] and the second time I was sharing the article title which presumably was the objectionable element), I am pretty sure both zuccings were the products of algorithms. And if not, this might reflect even worse on Facebook - if even humans can't detect sarcasm or article titles, good luck getting AI to do it.

On a related note: from 26 July Facebook won't allow third party applications to post to personal profiles, so I won't be able to use Buffer anymore.

The inability to queue posts and retarded zuccing combine to make a great incentive to scale back on my Facebook activity.

So I guess there's an upside after all!