Can you sue a ROBOT?

 Some forms of AI are so common that people don’t even consider them AI—Google, Siri, videogames, etc. Most people aren’t typically worried about liability when I talk about those AI. But when the topic changes to exciting forms of AI that are forthcoming—Google Car! Autonomous surgeons! Robot nannies!—liability immediately becomes a big concern. “Who would want a car that drives itself?! What if you hit someone?” “No one will make a robot babysitter—think about all the lawsuits!” “Don’t lawyers cause enough problems for doctors? You think they’re going to open themselves up to those lawsuits?”

These concerns, while valid, do not fully appreciate the changes that AI will introduce. Liability under the law is fundamentally concerned with assigning responsibility for wrongdoing. If someone is hurt through the actions of another, it is only reasonable that the second person be held responsible and that he or she help or pay the injured person. The assumption, of course, is that a person has done something and that a person will be responsible. The driver. The babysitter. The doctor. AI introduces another possibility. There is an accident, but all the people involved have acted responsibly and have not caused it. A machine, acting independently of people, caused it: Lt. Commander Data steers the Enterprise into an asteroid, Johnny Five from Short Circuit crashes a truck in Montana, etc....