You know what they say, comedy is suffering.
The whole world may be scrambling to further developments in artificial intelligence, but according to experts, one development we should hold off on is a sense of humor.
One of the biggest problems to this, says some scientists, is that AI will never be able to understand humor quite the same way as humans. “Artificial intelligence will never get jokes like humans do,” computational linguist and humor student at Texas A&M University-Commerce, Kiki Hempelmann, says. “In themselves, they have no need for humor. They miss completely context.”
This is a likely reason why humor should play a role in Turing Tests of the future. Turing Tests are considered the ultimate test for artificial intelligence, as they are used to determine if an outside evaluator can figure out that they are talking to a machine or a human.
“Teaching AI systems humor is dangerous because they may find it where it isn’t and they may use it where it’s inappropriate,” Hempelmann adds, going on to say that robot humor could get a person killed, just because it said an inappropriate joke at a wrong time, thinking it was funny.
Meanwhile, Columbia University computer scientist Alison Bishop says that humor and computers are incompatible since the latter looks for patterns while the former has to deviate from the pattern in order to deliver something funny, like a punchline. Bishop also adds, “I like to believe that there is something very innately human about what makes something funny.”