Almost everything on Facebook — the posts in your timeline, the ads, and features like speech recognition and automatic image tagging — is driven by artificial intelligence (AI).
And Facebook is just getting started. The social media company's new personal helper, M, can help you make dinner reservations, buy things off Amazon, and even recommend anniversary gifts for your spouse.
Behind the technology that powers M and other Facebook features is a special team led by Yann LeCun, an AI researcher and former New York University professor.
Since LeCun joined Facebook in 2013, the company has opened AI laboratories in California and in Paris, France, where LeCun was born.
LeCun has researched image recognition since the 1980s. He even helped revive an AI field of study called "deep learning," which now powers the self-learning abilities of Facebook's ad targeting, Apple's Siri, Amazon's recommendation services, and more.
Tech Insider emailed LeCun, director of the Facebook Artificial Intelligence Research team, to find out how AI is changing our lives, what most people get wrong about the field, and where the technology is headed.
You can read a version of that conversation below, which we've edited for length, style, and clarity.
TECH INSIDER: What do you think is the most impressive, real-world use of AI?
YANN LECUN: It changes quickly. I'd say the many new applications of image recognition used by Facebook, Google and others.
Some of these are apparent, like keyword-based image search and face recognition, some are acting behind the scenes, e.g. for content filtering.
TI: How did you become interested in AI?
YI: My dad is an engineer (retired now) and got me interested in designing and building things, particularly model airplanes and electronics.
I was fascinated by the mystery of intelligence as a kid, which got me interested in human evolution, epistemology and AI. Because I like building things, I taught myself electronics and programming in high school.
TI: Your field started as a way to study human intelligence by recreating it on computers. Does AI have to imitate how the human brain works?
YL: No, but AI is to the brain as airplanes are to birds. The details are different, but the underlying principles are the same. For airplanes and birds, the underlying principle is aerodynamics.
The question is, what is the equivalent of aerodynamics for intelligence?
YL: The short answer is: We have no idea. That's why it's very difficult to make a prediction as to when "human-level AI" will come about.
Right now, though, the main obstacle we face is how to get machines to learn in an unsupervised manner, like babies and animals do. Right now, the way we train machines is "supervised," a bit like when we show a picture book to a toddler and tell them the name of everything.
TI: What do you think are the biggest and most common misconceptions about AI?
(1) "AIs won't have emotions." They most likely will. Emotions are the effect low-level/instinctive drives and the anticipations of rewards.
(2) "If AIs have emotions, they will be the same as human emotions." There is no reason for AIs to have self-preservation instincts, jealousy, etc. But we can build into them altruism and other drives that will make them pleasant for humans to interact with them and be around them.
(3) Most AIs will be specialized and have no emotions. Your car's auto-pilot will just drive your car.
TI: Emotions play a big role in AI in popular culture. Do you have a favorite sci-fi depiction of AI?
YL: Most of them are terrible. But I like HAL in "2001: A Space Odyssey," not because it goes crazy and murderous, but because I watched the movie when I was nine years old, and I became fascinated by the very idea of AI.
I also like "Her." The robot child in the movie "A.I." is interesting. It's a harmless form of AI animated by the best human emotion: love.
TI: Are any of those sci-fi depictions close to being possible?
YL: Something like "Her" is not entirely implausible, but it's very far from what we can do today. We are decades away.
Almost all Hollywood depictions of AI and robots are implausible. AI either has to be emotionless, or if it does have emotions, they seem to be caricatures of the worst human drives and emotions (jealousy, greed, desire to dominate, becoming murderous when threatened, etc...).
AIs will not have these destructive "emotions" unless we build these emotions into them. I don't see why we would want to do that.
TI: What do you think this technology's greatest impact on society will be in the future?
YL: AI will transform society, but probably not more (relatively speaking) than past technological progress like cars, airplanes, telecommunications, regular computers, modern medicine, etc. Some jobs will disappear, others will be created as with every wave of technological progress.
There will be immediate benefits of AI like better medicine and reduced traffic accidents (due to self-driving cars). The overall wealth of societies will increase.
The questions are (1) how will societies adapt to share the benefits, (2) what human activities will become highly valued?
Number one is for the political process to solve. For number two: Uniquely human activities will become more valuable, for example artistic expression.