By David Dean on November 2, 2017

The Machines Are Coming

robotLong considered the domain of science fiction, recent years have seen numerous applications of artificial intelligence begin to enter the mainstream. Apple’s Siri and Amazon’s Alexa have brought artificially intelligent personal assistants into our pockets and our living rooms. Google Translate applies machine learning to language translation, with near-human accuracy.[1] IBM’s Watson is working with medical doctors in New York City to review patient records and recommend cancer treatment options. The first autonomous technology vehicles are already on the road.

Even the most conservative prognosticators expect that this trend will continue and grow, with artificial intelligence becoming an increasingly pervasive part of our lives. One expert has predicted that the automation revolution brought about by artificial intelligence will be 3000 times more disruptive than the Industrial Revolution.[2] To give just one example: 10 million AV enabled cars are expected on the road within the next 3 years, with major consequences for transportation, logistics, and insurance industries that can only be speculated about now.

Who Is Liable When a Machine Commits a Tort?

In May 2016, a fatal traffic accident in Florida inaugurated a new era in the realm of products liability. A Tesla Model S vehicle equipped with Tesla’s Autopilot technology collided with a semitrailer truck, resulting in the death of the Tesla driver. A National Transportation Safety Board investigation concluded that both human error and the limitations of the Autopilot technology contributed to cause the crash. However, the National Highway Traffic Safety Administration did not find a defect in the Autopilot system and declined to issue a recall.  In fact, the NHTSA praised the Autopilot technology for reducing the crash rate for Tesla vehicles by 40% since the system was first used.

The 2016 Tesla accident raises a fundamental question which will become ever more important as we entrust ever more artificially intelligent machines with ever-increasing autonomy in real world situations: who is liable when a machine commits or participates in a tort? The Tesla accident shows that it is no longer academic.  As we confer upon machines the ability to think, learn, and most importantly, to decide, courts will inevitably be forced to decide who is liable and how to apportion fault when one or more of the parties is not a human.

Will AI Usher in a New Liability Standard?

One commentator suggests that the rise of AI could fundamentally alter the law of product liability. Ryan Abbott, Professor of Law and Health Sciences at the University of Surrey School of Law argues in a forthcoming paper entitled “The Reasonable Computer: Disrupting the Paradigm of Tort Liability” that, “most injuries people cause are evaluated under a negligence standard where unreasonable conduct establishes liability. When computers cause the same injuries, however, a strict liability standard applies.” Abbott argues that the strict liability paradigm has negative repercussions when applied in the context of AI: “it discourages automation, because machines incur greater liability than people.”

Abbott has a more radical suggestion: that, as machines inevitably become more competent than humans, the “reasonable person” standard for negligence should be replaced by what might be called the “reasonable machine” standard. “This means that human defendants would no longer have their liability based on what a hypothetical, reasonable person would have done in their situation, but what a computer would have done. In time, as computers come to increasingly outperform people, this rule would mean that someone’s best efforts would no longer be sufficient to avoid liability.”

“At some point,” predicts Abbott, “computers will cause so little harm that the economics of negligence versus strict liability will be irrelevant. Autonomous computers will have become so ubiquitous that the constantly improving reasonable computer should be the benchmark for most or all areas of accident law.” Whether that will eventuate remains to be seen. But, one way or another, tort law will adapt to AI, as it did to cars and airplanes a century ago.

[1] Yonghui Wu, et al., Google’s Neural Machine Translation System: Bridging the Gap Between Human and Machine Translation (Sept. 26, 2016),