Welcome to the World of "Reasonable Robots"

           The development and adoption of the automobile changed forever personal injury law. The common law era of horses and buggies gave way to manufacturing, high speeds and unforeseen injuries. First came the car, then the law involving accidents with cars. Technology is always a step or two ahead of the law.            

           We’re on the cusp of a similar revolution today. This one involves artificial intelligence.            

           The gradual adoption of artificial intelligence and machine learning is transforming the world. Scholars are already trying to image the future, and forecasting/recommending ways the law might adapt.  

           Ryan Abbott’s, The Reasonable Robot: Artificial Intelligence and the Law, (Cambridge University Press, 2020), is a brief and highly readable look to the future.            

           Abbott brings a polymath’s perspective to the project. He is the possessor of a medical degree, a law degree, a Ph.D., and a degree in oriental medicine. He teaches law and health sciences at the University of Surrey and Medicine at UCLA. I suppose with all those degrees, he’s earned the right to write about intelligence.                

            He defines AI as an “algorithm or machine capable of completing tasks that would otherwise require cognition.” It’s a simple enough definition, but it begs the larger philosophic questions: What, exactly, is cognition? Is the phenomenon of mind simple a function of complexity? Given enough computing power, can a mind, an entity capable of self-reflection and goal oriented, arise from mere matter?    

         (A quick note for those of you new to the topic: What’s an algorithm? Simply put, it is a finite set of instructions or rules in a problem-solving device. Thinks of the formulae embedded in an Excel spreadsheet.)            

          These questions needn’t be answered for pragmatic purposes, and Abbott does not answer them. The fact is that we now see signs of AI all around us. We have created machines that can learn from experience. We have created neural networks than can accomplish tasks in opaque ways, meaning we see that the task was performed, but we are not quite sure how the machine reached the conclusion it did. AI is already transforming our world, whether we like it or not.                

            Abbott focuses on four discreet areas of the law: AI and employment-related taxes, personal injury law, patent law and the criminal law. He argues that as public policy evolves, policy makers ought to focus on laws, rules and regulations devoted to “AI legal neutrality,” by which he means that the law ought not to discriminate between AI and human behavior. It no longer suffices to regard a computer as a simple computing device. The digital world more and more resembles the human world: just ask Alexa, the next time she interrupts your dinner conversation with a question you thought you never asked.            

          The book is short and jargon free. I’ll leave you to explore what Abbott has to say about tax policy, personal injury and criminal law. (Spoiler: As a long-time criminal defense lawyer, I was disappointed in his treatment of the criminal law. It strikes me that AI possesses significant problems in the assessment of liability for harm to others. Whereas personal injury law has ready-made doctrines such a product liability theories that can at least provide a launching pad for application of new rules to machines, the criminal law is more hide-bound, adhering to such ancient tropes as free-will, autonomy and personal responsibility – concepts that are difficult to apply to machines. I’ve written on this page before about AI and criminal law, and will do so again.)          

          But consider the following: Should an AI device be capable of being granted a patent? In other words, should artificial inventors have the same rights as human inventors? Net neutrality suggests that the answer is yes.            

          Patent law is grounded in a sense of social utility. We give to inventors exclusive rights to the licensure of their work because doing so encourages and incentivizes innovation. But we we don’t give out these exclusive rights at the drop of a hat: that, too, would discourage the exchange of information and yield a less productive society. Under current law, only humans can be inventors. That’s the way federal patent law is written, and that’s how the courts interpret it.            

         But the truth is that machines make discoveries. Consider the human genome, the 3 billion or so base pairs of nucleotides that make us who we are. We’ve decoded the human genome, and located also sorts of places on the genome where genetic work is done that makes who we are. We can select for desirable features. We can also engage in designer medicine, where machine learning matches potentially life-saving therapies to the particular genome of an individual patient. Who “discovers” that?  Consider also intellectual property: AI can write music, compose a poem, write a short story or even a novel. Why no copyright?

            (The answer is not as simple as the owner of the machine. Often, especially in open-source coding and cloud computing, it’s not clear who, among the commons, owns what.)            

            Abbot was part of a team that, in 2019, as part of something called The Artificial Inventor Project, offered to patentability an invention created solely by an AI device. (You can learn more about the project at www.artificialinventor.com) The patent claims were ultimately rejected because although the machine generating the inventions did so in a manner indistinguishable in formal characteristics from that of a human, the machine was, well, not human. Isn’t that a distinction without difference, Abbott wonders, and doesn’t the failure to offer patentability impoverish us all by disincentivizing those who control AI to set their creations free to flourish and create things from which we can all benefit?            

          Why not create a world in which AI can receive patents?             We’re not yet in an era of General Artificial Intelligence, that is, a world in which a device can do all the things a human can do. But we do know that artificial intelligence can operate automobiles in way that are safer than a human, potentially saving lives. Narrow Artificial Intelligence, that is, algorithms focused on particular tasks, exceed human performance in many areas. There’s a brave new world right around the corner.  

          Abbot doesn't have all the answers. Who does? But's he's raising the right questions.          

          We can’t stop the new world from arriving. We can only prepare to harness it for human flourishing. Abbott’s volume is a welcome addition to the growing literature of ebate about how best to prepare.                        


  • No comments yet

Add a Comment

Display with comment:
Won't show with comment:
How many sides does a triangle have?
*Comment must be approved and then will show on page.
© Norm Pattis is represented by Elite Lawyer Management, managing agents for Exceptional American Lawyers
Media & Speaker booking [hidden email]