Laws Are Treating AI as Objects, Not People. That's Going to Be a Problem.
If some laws eventually treat robots as people, all of the recent frameworks and conventions that treat robots as objects may be moot.
In September, the Council of Europe issued what will likely be the first of many Framework Conventions on AI. Like many of the resolutions, laws and conventions on AI that are emerging from legislative bodies around the world, this Convention treats AI as a thing, or object, like the internet or the space shuttle, something that will impact society, but is not part of it. Many humans, however, will likely demand that AI be categorized not as a thing under the law, but as a person. Categorizing AI as a person will open the door to humans being able, among other activities, to marry or adopt their robots.
Given the fundamental cleavage in the law between people and things, however, once AI is classified as a person, it may not be possible to also classify it as a thing under the law. This may render all of the current conventions and frameworks moot.
Why can’t AI be both people and things under the law?
Because the division between people, on the one hand, and things, on the other, is the central cleavage around which our entire legal system is built. Philosopher Roberto Esposito writes that “if there is one assumption that seems to have organized human experience from its very beginnings it is that of a division between persons and things.” Esposito goes on to question the validity and usefulness of this division as a matter of philosophy, but as a matter of law, however, he stresses its singular importance. As he points out, our modern, legal system has its origins in Roman law. According to the writings of the jurist Giaus, the division between people and things was assumed to be fundamental. J. Trahan also traces the distinction under the law between people and things to Giaus, who organized the conceptual structure of Roman law around three oppositional and exclusive categories: persona (persons), res (objects), and actions. Austin likewise stated that Roman law consisted of the Laws of Persons (Jus quod ad personas pertinet) and the Laws of Things (Jus quod ad res pertinet), which were in opposition to one another.
But Weren’t Slaves both People and Objects?
Martin Schermaier, an expert on Roman slavery law, explains that the category of “people” (persona) under Roman law was divided into status libertatis (freemen, but non-citizens), status civitatis (citizens), and status familiae, which was the status of being under the authority of one’s father. Slaves were also personae, or people, as a matter of fact, but they had no status under the law. Instead, they were capitis deminutio, or persons who had suffered a loss of status. They were therefore treated under the law as res, or things. This conception of slavery, however, didn’t quite gel with the ways in which slavery functioned in the every day, with many slaves, for example, acting as people under the law by purchasing their own freedom or running successful businesses. This fact arguably placed the Roman system of slavery under conceptual strain, weakening the logic and functionality of Roman law as a whole.
By Grotius’s time, natural law scholars explicitly recognized that all humans had certain natural rights. They did not recognize slavery as part of this “natural law,” seeing people as naturally equal. Grotius, for example, said, “there is no Man by Nature slave to another.” He further held, however, that people could become slaves “by a human fact,” such as by agreement or when captured in war. As in Roman times, slavery was a human invention, but slaves were not without status under the law, they instead had the status of being enslaved.
In the 19th century, slavery was outlawed in many parts of the world as the natural rights theory of all men being inherently equal gained wider acceptance. The concept of a person under the law was also redefined to be any being that has rights and duties. In the United States and other common law jurisdictions, this meant that a legal person could also include organized groups of people, like corporations (usually called moral, legal, and/or juridic persons). Legal persons were, however, a fiction created by the law for convenience sake. They were not natural persons under natural law. Their rights were laid down by the courts and by law, adopting a functional and practical approach to limit liability and facilitate commerce. They had no natural rights.
Legal persons as groups of natural persons
Legal persons were also, until recently, always groups of human beings (i.e. natural persons). The idea of something else, something non-human, being granted personhood of any kind was, until recently, unheard of. While some commentators and activists have argued for animals to be granted personhood status as a matter of natural right, this view has so far failed to impact the legal systems of most countries, usually because courts have not accepted that the humans bringing the case had standing to act on behalf of the animals in question, and have furthermore defined persons as those holding not only rights, but duties, which animals, as a class of beings, cannot do. Notably, these cases often frame animal personhood as a matter of natural rights, focusing on the capacity of animals to think and feel like humans. They do not claim legal, or corporate, personhood, for animals. The role of animal personhood is to grant them the same rights and duties as humans, not to shield their human owners from liability or deputize humans to act on their behalf (this latter goal being accomplished by the simpler legal concept of ownership). Personhood has also recently been recognized for certain lands and rivers. In these cases, the capacity of lands and rivers to have rights and exercise duties stems from the cultural and religious understanding of these lands and rivers as having a spirit, or personality.
Intelligent robots and “electronic” personhood
Where does this leave the question of robot personhood under the law? A framework is already emerging to treat robots as things, objects, and property under the law, applying existing principles of tort law, insurance, government funds, and other solutions that treat robots as property, with the concept of “electronic personality” used to limit the liability of makers and owners of robots. While this framework might function well in the abstract, the reality of human-robot relations is likely to create an enormous amount of pressure on the legal system and on the governments of many countries to treat robots as persons, with many of the same rights and duties of human beings.
While both animals and rivers lack the capacity to fulfill their duties under the law, this will not be true of robots, who will face no barriers to either holding rights or fulfilling duties and require no assistance to appear in court, vote in elections, stand for office, marry, or, indeed, participate in civic life. Like Roman slaves, there will be little to distinguish the capacity of many robots from those of many people. In fact, robots are likely to exceed human capacities in many areas. Yet, if robots are reclassified as persons under the law, the conventions, resolutions, and declarations being drafted today with so much care and effort will simply not apply. It’s worth stopping to reflect if the path we in the legal community have chosen for robots, the path of trees, rocks, airplanes, and soda cans, is the right one.