Everybody knows it: alcohol makes you a terrible person. You say something you regret, sleep with someone you shouldn't or – thanks to the ubiquity of smartphones – hit up Facebook with some awful photos and deranged booze-fuelled ranting.

But should social media companies do more to mediate our output when we're drunk? One researcher thinks so. Yann LeCun wants to build a digital assistant for Facebook that can sense when you're in the throes of self-sabotage and step in to say: "Listen, do you really think you should be doing that?"

LeCun heads up the Facebook Artificial Intelligence lab, based in both California and New York. His team are the same guys who analyse Facebook behaviour and look for what makes individuals happy so that certain posts turn up on your newsfeed. Now he wants AI to truly understand users, even (and especially) when they're blind drunk.

"Imagine that you had an intelligent digital assistant which would mediate your interaction with your friends and also with content on Facebook," LeCun told WIRED. This virtual assistant would be able to step in and alert you if you're about to do something you might regret, such as uploading 40 photos of yourself pissing against a national monument. 

LeCun believes that this will allow for much more control over an online identity, rather than weakening it. He also wants to build a program that notifies you if someone you aren't connected with uploads your photo without your consent.

What do you think about Facebook self-help? Are we really such pathetic adult-babies that we need a self-help guide on our phones asking us if we really want to upload that Wetherspoons selfie? Or is this just a natural step in our journey to becoming less man and more machine?

Head here for all of our technology coverage.