Campus Review Volume 28 Issue 12 December 2018 | Page 21

policy & reform campusreview.com.au The morality of driverless cars The public should have a say on how autonomous vehicles make life-threatening decisions, scientist argues. By Loren Smith A t least 11 auto manufacturers, including Toyota, Audi and Tesla, are putting driverless car plans in motion. With the soon-to-be robot drivers comes the need to program ethics into them. Think about a situation like the trolley problem, where a driver must choose between intervening and killing one person to save five, or doing nothing and allowing those five people to die. But who should make these life‑or‑death decisions? A Syrian-Australian MIT scientist and his team are trying to convince tech companies and policymakers that the public should have a say. Associate professor of media arts and sciences Iyad Rahwan and his colleagues have compiled a database, called the Moral Machine, of 40 million individual decisions to make their case. The responses derive from the citizens of 233 countries and territories. The research shows that humans have some core, shared morals. When it comes to who dies and who should be spared, people privilege humans over animals, the young over the old, and saving more lives than fewer. Yet across the globe, certain opinions differ. For example, the researchers claim that residents of Central America and France feel similarly about prioritising the lives of women and those with athletic 10 prowess. Also, for those from more economically unequal countries, like Russia, social status appears to be an important determinant of who should live. The researchers’ reflections on people’s responses were compiled in an article published in Nature. Other researchers, including some from Australia, also weighed in on the findings. Distinguished Professor Mary-Anne Williams, director of disruptive innovation at the Office of the Provost at UTS, wondered about the trickle‑down effects of driverless car morality programming. “In order to minimise liability, car companies may design cars that slow down in wealthy neighbourhoods, or that kill humans rather than cause more expensive serious injuries,” she pondered. Williams added that car morals may be difficult to apply in practice, as accidents often aren’t as clear cut as choosing between the death of one person or the deaths of many. Hussein Dia wondered whether human thoughts, crowd-sourced or not, should dictate machine ethics. The associate professor in transport engineering at the Swinburne University of Technology noted that the only existing proposed framework for this does not discriminate between people dying on the basis of personal characteristics. “This clearly clashes with the strong preference for sparing the young [such as children] that was assessed through the Moral Machine,” he said. Professor Toby Walsh, perhaps the most well known Australian AI specialist, added that people‘s stated intentions don’t always tally with their behaviour. Therefore, their responses in this study should be treated with caution. The research leader of the Optimisation Research Group at Data61 also raised a potential issue with the study’s design, throwing its findings into further doubt: “I completed their survey and deliberately tried to see what happened if I killed as many people as possible. As far as I know, they didn’t filter out my killer results.” Regardless, like Dia, Walsh questioned whether human morals should be applied to machines at all. “We should hold machines to higher ethical standards than humans for many reasons: because we can, because this is the only way humans will trust them, because they have none of our human weaknesses, and because they will sense the world more precisely and respond more quickly than humans possibly can.” Associate Professor Jay Katupitiya, from the UNSW School of Mechanical and Manufacturing Engineering, extended this viewpoint: “In my opinion, programming these intentions is more immoral than not.” Unlike Walsh, however, he believes the morality of driverless cars isn’t the biggest road safety issue; human-driven cars are. As such, he thinks we shouldn’t become too obsessed with “rare ‘dilemma’ cases” like the trolley problem. Rather, we should focus on the “real, everyday safety gains” this technology can offer. Yet Katupitiya is something of a lone voice in this debate. Most academics in the field seem deeply concerned with this morality issue – which extends to AI more broadly. Perhaps when it comes to life or death, the resolution of moral dilemmas has added urgency. ■