When it comes to artificial intelligence (AI) and machine learning the two things that make a difference are: A. The direct experience you have in solving computing cycle problems, data classification issues, and data trustworthiness questions and B. The number of world-class, qualified engineers you have at hand.
Facebook has neither of those. Its direct experience of AI is recent which means it doesn't yet go deep and it is struggling to find and keep top-notch engineers who could help the company aggressively tackle the issue.
So, what does it do? It lies. :) (check it out: https://goo.gl/0FKibE) It releases an 'AI' product it calls "M: which is actually powered by humans trained to pretend to be an AI. At a time when everyone else is busy going the other way (using AI and training it to try and be human) the reasons Facebook has chosen this path are simple: The company knows that the future of machine/human interaction will be governed at the UI by contact with an artificial intelligence. It doesn't have one. Probably not even close. But if it cannot yet provide some experience of the interface so that its billion plus users can become acquainted with it, trying to play catch-up later will be a difficult task.
That much we get. It's no big deal. But to actively lie to everyone about it shows that despite its troubles with its reputation Facebook continues to behave like we are stupid and it knows best. Sadly it also means that the company has no idea how to develop trust and how to use it.
For a deeper context against which all this is happening you should read 's piece on chatbots: http://goo.gl/Sn9VaB.