You may not find the antics of R2-D2 terribly charming, but apparently a bit of bumbling can make a robot much more appealing to the humans that need to interact with it. This may seem counterintuitive, since any robot that’s tasked with assembling our airplanes or fighting fires should perform those tasks as accurately as possible to help keep us safe. However, robots performing more socially-oriented functions, like helping you check in for a flight, are more likely to get on your good side if they make an error or two (and correct it.) This is great news for roboticists, since we’re a long way off from any robot that can handle every social interaction perfectly… not unlike actual humans.
A study from the University of Salzburg, Austria, asked people to perform various tasks with a robot assistant. The tasks weren’t the focus of the study though, as the real variable was how the robot behaved along the way. Some robots performed the tasks as smoothly and flawlessly as possible, and were were generally rated by their human partners as being very anthropomorphic and intelligent. Other robots were programmed to make mistakes from time to time, and while they weren’t rated as being so intelligent, they were nonetheless the most likable robot to partner with. This is important, because people will be more willing to work with an automaton on any project or transaction if they can find it likable in some way.
Our fondness for flaws
Researchers believe that this is a case of what’s known as the Pratfall Effect. It has been demonstrated experimentally by showing people video of a peer successfully answering questions in a game show setting. After getting 92 percent of the questions right, some viewers see this person spill their coffee in their lap, while others don’t. Everyone can agree that this is a highly competent person, which is good, but the people who saw the spilled coffee also find the person to be very likable. Small flaws make a person, or robot, feel relatable and “human.”
Mistakes only make people, or robots, more attractive if they’re already seen as competent though. So a robot can’t do everything wrong and expect to win anyone over. However, perfection isn’t a practical goal, so researchers propose that robots be designed to take advantage of the Pratfall Effect when mistakes are inevitably made. If a robot can read social cues from humans well enough, it may be able to better scrutinize it’s own behavior for mistakes to correct. Correcting the mistake would then show competency, but also “humanize” the robot in a way that would be otherwise hard to plan for.
Source: Why Humans Find Faulty Robots More Likeable, Scienmag