An MIT project is learning to predict how humans will interact upon first meeting based on patterns from over 600 hours of TV footage

MIT’s CSAIL team has created an algorithm that can predict how two humans will greet each other, based on hours of footage from popular TV shows like The Office and Desperate Housewives.

The algorithm can predict whether two people will hug, kiss, shake hands or high five, mimicking humans’ innate ability to predict interactions based on experience.

The algorithm uses deep learning technology, a field of AI that uses “neural networks” for computers to comb through data and find patterns. The system studied over 600 hours of video footage from TV shows to learn how to predict human behavior. “Humans automatically learn to anticipate actions through experience, which is what made us interested in trying to imbue computers with the same sort of common sense,” CSAIL PhD student Carl Vondrick explains, “We wanted to show that just by watching large amounts of video, computers can gain enough knowledge to consistently make predictions about their surroundings.”

This content is available for Basic Members.
Already a member, log in