A University at Buffalo professor has taken part in a study with a computer system that reads emotions better than humans.
UB Communication professor Mark Frank says the study involved showing videos of a person in real pain and in fake pain, and asking humans which is which.
"The human, on the average, they were getting somewhere around chance accuracy about 52 percent," Frank explained.
"Whereas, the machine learning techniques does considerably better, closer to 85 percent accuracy."
According to Frank, that's because a lot of what people know about emotions and behavior is false.
"The second thing is often these behaviors are subtle," Frank said.
"They're often quick, less than half a second, less than quarter second in duration, and people don't see them."
And he says the machine isn't caught up the biases that humans have.
"So you put all those things together and then you see why the machine will out perform the human in this judgment task."
Some of the potential uses include tracking people recovering from depression.
"As people are getting better, the types of smiles they show are different, and maybe this can be a helpful adjunct to a therapist," Frank said.
Frank believes some of the findings could be applied in "a law-enforcement, counter-terrorism, counter intelligence type of environment."
But even with advances in technology Frank says a human that understands behavior will always have to be involved.