Robots can push people to engage in riskier behavior, according to recent research, underscoring how technology can shape human decision-making and risk-taking.

The study, published in Cyberpsychology, Behavior and Social Networking in November, examined the impact a robot could have in a risk-taking scenario. Participants took the Balloon Analogue Risk Task (BART), a computerized decision-making task that asked them to inflate a balloon on the screen by pressing a keyboard space bar. 

Each press of the space bar would slightly inflate the balloon and also add a penny to the participant’s “temporary money bank.” If the balloon popped, the person would lose all the money in their bank, though they could also “cash-in” their bank and move on to the next balloon before the current one popped.

The research included 180 participants divided into three groups. The first group had no robot present, while the second and third groups featured Pepper, a humanoid robot from SoftBank Robotics. Pepper was silent in the second group, and in the third group provided instruction as well as encouraging questions, such as “Why did you stop?”

Individuals who were encouraged by the robot tended to take more risks, pressing the space bar 23% more frequently than the robotless group and 22% more frequently than the silent robot group. These participants would accordingly blow up their balloons 38% more frequently than participants in the other two groups.

There was not a noticeable difference between participants in the robotless and silent robot groups in terms of their risk-taking as assessed by the BART. These groups both showed a tendency to scale back their risk-taking behavior, pressing the space bar less frequently, following a balloon explosion.

Yaniv Hanoch, an associate professor at the University of Southampton and an author of the study, said that while the researchers had predicted the outcome, they were still somewhat surprised by their findings. 

“There’s no other research that has shown that robots can actually influence humans to that degree, let alone in risk-taking,” Hanoch said. “We’re only starting to realize how much robots can influence human behavior.”

Hanoch noted that the study’s results may “open a Pandora’s Box” with regard to artificial intelligence mechanisms and how they can influence human behavior, given that there are already many concerns about robotics and AI technology, but he noted that the possibilities indicated by the study don’t “necessarily have to be bad.”

“We illustrated the fact that [AI] can increase risks, so there are maybe professions where that could be beneficial,” Hanoch said. “If you are in a profession that needs high risk-taking, where an employer or an organization wants you to take higher risks, possibly it would be beneficial … Every new technology produces anxiety and fears in us, because we do not know where people are going to take it, but it also should instill some hope and ambition in us, in the sense that it can lead to positive things.”

Hanoch said that the research team will be exploring whether or not the same kind of mechanism can be used to reduce risky behaviors in additional studies. The study used a financial incentive for button-pressing in the BART, so further research may explore how robotic encouragement can reduce risk-taking by demographics such as gamblers. Other groups, such as teenagers, may also be focused on in later research, as this study examined adults.

“Teenagers are a really important group where sometimes we want to reduce their risk-taking activity – driving, especially,” Hanoch said. “Do we have a mechanism that we can install in a car, for example, that could reduce risky driving behavior among teenagers? Among that age group, traffic accidents are one of the leading causes of death. Can we install something in a car and reduce risky driving behavior? That would be quite promising.”

The key to the study is that it showed how robots can indeed influence human behavior, Hanoch said, and highlighted that further research is needed to explore the potential of that influence.

“It’s really in its infancy, this kind of research,” Hanoch said. “The sky’s the limit, more or less.”

The article, “The Robot Made Me Do It: Human-Robot Interaction and Risk-Taking Behavior,” was published Nov. 18, 2020 in Cyberpsychology, Behavior, and Social Networking. It was authored by Yaniv Hanoch of the University of Southampton, Francesco Arvizzigno of the University of Bari Aldo Moro, Daniel Hernandez Garcia of Manchester University, Sue Denham of Bournemouth University, Tony Belpaeme of Ghent University, and Michaela Gummerum of the University of Warwick.