Italian scientists tested the AI thinking processes on a robot called Pepper before making a decision. It turned out that when robots think aloud, they do their job better.
"There is a connection between the inner dialogue and the subconscious in humans, so we wanted to see what it looks like in robots," said Arianna Pipitone, who supervises the science experiment, from the University of Palermo.
To this end, the Guardian explained, researchers have programmed a robot called Pepper that can vocalize their thought processes so that they can hear how it affects their actions. Pepper was asked to place a napkin on a set table contrary to the rules of etiquette. The robot, while talking to itself, concluded that the person who had given it to him had made a mistake, and wondered what to do next. When his user said that he would like his wishes to be done, Pepper commented, "This situation is sad for me, I would never break the rules, but I can't let him down, so I'll do as he wants." Then he placed the napkin in the wrong place.
After comparing how Pepper works, when he does internal dialogue and when he doesn't, researchers concluded that he is more efficient when he talks to himself.
The study, published in iScience, concludes that this function in artificial intelligence can be very useful when robots need to collaborate with people, for example in caring for dependent people. "But also annoying when simple actions are required, such as turning off the light. (...) But maybe one day a robot will be created that will spontaneously generate speech," said Pipitone. (PAP)