Facebook head of AI wants to stop you from posting things you may later regret

Facebook artificial intelligence (AI) could someday stop you in your tracks as you drunkenly post photos of yourself at a party for all the world (and your mom, and your boss) to see.

Or, at least, that's what Director of AI Research at Facebook Yann LeCun says he's hoping to get working, using a form of AI called "deep learning".

LeCun, the New York University researcher and machine-learning guru who now oversees Facebook's year-old Artificial Intelligence Research lab, told Wired that one of the lab's next goals is to craft a virtual assistant who can recognize unwise posting behavior and then virtually tap us on the shoulder to say something along these lines:

Uh, this is being posted publicly. Are you sure you want your boss and your mother to see this?

Deep learning has also been called a rebranding of neural networks, where the basic idea is to build machines that function like the human brain.

Whereas past technologies such as Google search have tried to do that by rapidly analyzing huge amounts of data, researchers such as LeCun are instead aiming to build massive neural networks that mimic our brains. 

Not that we can claim to understand exactly how our own brains work, as pointed out recently by Michael I. Jordan, one of the world's most respected authorities on machine learning.

But Facebook's deep learning algorithms are already at a point where they can examine our overall Facebook behaviour - a capability that it relies on as it tries to figure out what News Feed content we're likely to click on versus which content prompts us to lunge for the "hide" button.

According to Wired, those algorithms will soon also analyze the text we type into status posts, automatically suggesting relevant hashtags. 

LeCun and his team plan to dive yet deeper still into understanding Facebook data in even more complex ways, to the point where the network can point us in directions we might not have gone on our own.

Wired quotes LeCun:

Imagine that you had an intelligent digital assistant which would mediate your interaction with your friends, and also with content on Facebook.

If you don't cotton on to the idea of your face being recognized and filed away somewhere in Facebook Dataland, and/or if you're not too keen on the notion that technology's going to morph into a nanny state that urges you away from making truly embarrassing Facebook posts, LeCun assures us that this is about more control over online identity, not less.

For example, he envisions a Facebook that's on alert, keeping an eye out for your data, and could thus give you a heads-up when, for example, someone you don’t know posts your photo without your say-so:

You will have a single point of contact to mediate your interaction but also to protect your private information.

Will the protection be worth whatever tradeoffs are implied by a more powerful, approaching-omniscient Facebook?

Your thoughts are welcome in the comments section below.