When it’s person versus machine, people don’t like to perform worse than their robot counterparts. And when they do, defeat hurts their confidence, a new Cornell study found.
Past research on human-robot interaction have largely focused on collaboration instead of competition, making this study a first in the field, according to Prof. Guy Hoffman, mechanical and aerospace engineering, one of the researchers.
Modeled after a similar economics experiment where two humans competed against each other, the study aimed to examine “loss aversion,” a behavioral economics concept that suggests people are more willing to avoid loss than gain awards. Loss aversion occurs when people are compared to a specific reference point — in this study, the robot.
Human participants of this study competed against a robot in a repetitive task that involved counting the number of times the letter “G” appeared in a string of characters, and then placing a block in the bin corresponding to that number. Their robot competitor performed at a randomly selected skill level. After each round, the winner of the task was awarded a random amount of money.
By using a robot, the researchers could manipulate the skill level of the robot, establishing a precise reference point to measure people’s reaction when robots performed better or worse than them, according to Prof. Ori Heffetz, economics, another researcher of the study.
“It is hard to find a laboratory task in which humans and robots can perform roughly at the same level, and that you can finely control. We also wanted [the task] to be monotonous, and uninteresting,” Hoffman and Kshirsagar said.
The study observed a small discouragement effect of the robot’s performance on the human participant’s performance: In general, participants liked low-performing competitor robots over high-performing robots. The robot’s performance also affected participants’ perception of their own abilities.
Opposite from the researchers’ expectations, the findings also show that the size of the prize awarded did not significantly affect people’s performance.
Moving forward, the researchers are also interested in understanding more about loss aversion and reference points.
“We people change our reference points all the time: We compare ourselves to our siblings, our classmates, our neighbors, or co-workers. Our research goal is to better understand these processes,” Heffertz told The Sun.