Researchers tortured robots to test the limits of human empathy

Back in 2015, a cheerful three-foot-tall robot armed with pool noodles embarked on what appeared to be a straightforward journey. HitchBOT, as it was called, aimed to hitchhike across the continental United States with the help of kind strangers. However, its adventure came to an abrupt halt after covering just 300 miles. Two weeks into the trip, HitchBOT was discovered abandoned on the streets of Philadelphia, its head severed and its spaghetti arms torn from its bucket-shaped body.

“It was quite a setback, and we didn’t really anticipate it,” shared hitchBOT co-creator Frauke Zeller with CNN at the time.

The unfortunate fate of hitchBOT is not an isolated incident. Throughout the years, humans have taken pleasure in mistreating robots by kicking, punching, tripping, crushing, and running over them. While this behavior may seem amusing, it could raise concerns as humanoid robots are being developed to collaborate with humans in manufacturing settings. Research suggests that people may show more compassion towards robots and treat them more gently if they exhibit human-like signs of pain. In essence, HitchBOT might have had a better outcome if it had been programmed to plead for mercy.

Humans feel guilty when robots cry

Marieke Wieringa, a researcher at Radboud University Nijmegen, conducted a series of experiments to observe how individuals reacted when instructed to shake a test robot forcefully. In some scenarios, participants shook the robot without any response, while in others, the robot emitted a pitiful crying sound or displayed enlarged “eyes” to convey sadness. The study revealed that participants felt more remorseful when the robot exhibited emotion-like reactions. When given the choice between shaking the robot or completing a mundane task, participants opted to proceed with the task when the robot cried out.

Wieringa stated, “Most people had no problem shaking a silent robot, but as soon as the robot began to make pitiful sounds, they chose to do the boring task instead.” Wieringa will be presenting this research as part of her PhD thesis at Radboud University in November.

These findings build upon previous studies indicating that individuals tend to treat robots more kindly when they display human-like characteristics. For instance, participants in one study were less inclined to harm a robot with a hammer if the robot had a backstory describing its personality and experiences. Additionally, test subjects showed friendlier behavior towards humanoid-shaped robots after experiencing the machine’s perspective through a VR headset. Other research suggests that humans are more likely to empathize with and trust robots that appear capable of recognizing their own emotions.

“If a robot can feign emotional distress, people feel more guilty when mistreating the robot,” Wieringa added.

The many ways humans have abused robots

The tendency to take out frustrations on inanimate objects has a long history among humans. From parking meters to vending machines to malfunctioning toaster ovens, people have exhibited a peculiar inclination to attribute human-like animosity to everyday items, a phenomenon described by writer Paul Hellweg as “resentalism.” Before the modern concept of robots, individuals were seen attacking parking meters and aggressively shaking vending machines. As technology advanced, so did the methods of destroying machines. This penchant for robot destruction was epitomized in the popular television show Battle Bots during the 2000s, where audiences cheered as hastily constructed robots were repeatedly sliced, shredded, and set on fire before a jubilant crowd.

With the prevalence of consumer-grade robots in the real world, some of these exuberant attacks are now occurring on city streets. Autonomous vehicles from Waymo and Cruise have been targets of vandalism, with tires slashed and vehicles set on fire. In San Francisco, residents reportedly toppled an egg-shaped Knightscope K9 patrol robot and defaced it with excrement after it was deployed by a local animal shelter to monitor unhoused individuals. Knightscope previously reported an incident where an intruder fleeing a healthcare facility intentionally ran over one of their robots with a vehicle. Food delivery robots operating in various cities have also been knocked over and vandalized. More recently, an AI-powered sex robot, valued at around $3,000, displayed at a tech fair in Austria had to undergo repairs after being reportedly left in a heavily soiled condition by event attendees.

One of the most notable cases of sustained robot abuse comes from Boston Dynamics, now owned by Hyundai. The company is renowned for creating some of the most advanced quadruped and bipedal robots globally, subjecting them to numerous attacks in the process. Videos circulating on YouTube show Boston Dynamics engineers kicking their Spot robot and harassing the Atlas humanoid robot with weighted medicine balls and a hockey stick.

Research into the reasons behind the enjoyment of abusing robots has yielded mixed results. In cases involving autonomous vehicles and factory robots, these machines can serve as a reminder of potential job displacement or other economic challenges stemming from increased automation. On the other hand, researchers like Agnieszka Wykowska, a Cognitive Neuroscientist at the Italian Institute of Technology, suggest that the lack of human traits in machines may trigger a peculiar form of tribal response known as anthroposophy.

“You have an agent, the robot, that is in a different category than humans,” Wykowska explained in a 2019 interview with the New York Times. “So you probably very easily engage in this psychological mechanism of social ostracism because it’s an out-group member. That’s something to discuss: the dehumanization of robots even though they’re not humans.”

As robots become more integrated into public life, our inclination to interact with them in various ways could become more complex. Companies like Figure and Tesla envision a future where humanoid robots work alongside humans in factories, assist with household chores, and potentially care for children. While these predictions remain largely theoretical, the success or failure of these machines may hinge, in part, on manipulating human psychology to evoke empathy towards robots akin to how we would feel towards a person.