Law does not cope with issue of AI as legal person
CAN A COMPUTER SYSTEM BUY AND SELL PROPERTY, EITHER IN THE ORDINARY UNIVERSE OR IN THE METAVERSE?
The biggest question facing the law worldwide in the next few years is how to recognise self-generative AI thinking systems as legal persons.
I use “artificial intelligence” in contrast to natural person intelligence, not because there will be anything artificial about what AI does autonomously in the near future.
Private law governs the rights and duties between persons who may be natural persons, or juristic persons such as corporations. Public law determines the rights and duties between the state as the sovereign power and all persons.
According to legal science, human beings, and the corporations through which they act, govern their independent actions as legal subjects and they have rights, duties and capacities. As the law developed, it was accepted that the law also protects the products of a legal subject’s mind, for example a painting or a book.
What happens when a computer system with selfgenerating intelligence is able to think for itself and create AI-generated content without acting on behalf of or under the instructions of human intelligence? These are not tools creating content. They are active participants making their own decisions based on data and acting on them autonomously — for example, buying and selling shares.
The question has already arisen in relation to self-drive vehicles and crewless autonomous cargo ships but those are the agents of, and programmed by, human beings. What happens when self-drive modes of transport can make autonomous decisions? Who gets the driver ’ s licence?
One of the basic principles of the law of persons is the requirement that one should not cause harm to another. If a selfgenerating computer system causes damage negligently or even willfully who do you sue for damages, and who do you put in jail for a crime? Or do you just close them down (if you can) and sell their intellectual property to pay the damages, if it is worth anything?
The other side of the coin is that a legal subject usually has rights and obligations in relation to a legal object. Can a computer system buy and sell property, either in the ordinary universe or in the metaverse? The whole principle governing the rights and obligations of persons is to hold people accountable for their actions and transactions.
How happy will we be to have a computer system as the lessor of our office premises?
The law does not cope with the issue. Since 1957 the Interpretation Act has defined a “person” for the purposes of any statute as including certain government bodies, incorporated companies and other incorporated and unincorporated “bodies of persons”. The word “person” is clearly used in the sense of a natural person.
The constitution refers to the rights of persons throughout, but relates to a natural or juristic person. Does a right to life include electronic life? If computer systems can make a cup of coffee, or anything, merely by watching other people do so, who bears any adverse consequences?
Consumer legislation such as the Consumer Protection Act does no better. What will protect the computer-wronged consumer, natural or artificial? The UK supreme court recently held that AI cannot be an inventor under their patent laws and cannot be granted patents for inventions. Will we benefit by having the inventions of computer systems unrecognised as protectable rights?
These are not just musings of idle minds. The self-generating intelligence of computer systems is developing faster than most natural intelligence can cope with. We have to revise basic concepts of law which go back to Roman times to protect persons and their interactions, including autonomous persons which do not act through human agents.
The consequences of not doing so will be far-reaching.