Ethics, humans, and AI
By Raamesh Gowri Raghavan and Sukant Khurana
To begin with, ethics, morality and the law are distinct entities, even though any system of law is based on the ethics of those enforcing or abiding by the law. Ethics are usually absolute, while morality, which is often enforced by religious and social forces, is relativistic. Thus while it is considered unethical in all circumstances to kill a human being, it may not be considered immoral in certain circumstances by many societies, such as in war, or religiously-sanctioned murder (of an apostate, for example). Law interacts with ethics and morality in similar ways; thus killing a person declared an ‘enemy of the state’, such as a person convicted of ‘treason’, or an armed citizen of another country designated as an ‘enemy soldier’ is not deemed an illegal murder. On the other hand, euthanasia is illegal under the laws of many countries, including India, although some religions, such as Jainism, may not consider it immoral.
But whose conduct may be deemed to be ethical or otherwise, immoral or illegal? For this, we come to two more concepts of philosophy that have been debated from time immemorial: personhood and agency.
Personhood is the recognition that an entities is entitled to some rights, and simultaneously subject to liabilities; as given by a state, religious body or social customs. For the purpose of this paper, we shall restrict ourselves to considering personhood under the law only, for the other forms are not enforceable in modern societies. The law of almost all countries recognises two forms of personhood, natural and juridical.
In almost all countries today, full natural personhood is extended to all human beings above the legal age. Such persons, whether they be citizens or not, are entitled to certain rights, responsible for certain duties, and liable for acts of omission or commission. Those below the legal age of majority may not receive complete personhood, and therefore restricted from exercising certain rights (such as voting) and holding liability for criminal acts (hence minors are not liable to face the full fury of the law).
Juridical personhood is granted to non-human entities that are to be treated as ‘persons’ for the use of the law; this is always extended to sovereign states, government agencies, companies, NGOs, associations etc that execute financial transactions, and may be prosecuted in a court of law. This is the basic tenet of law that allows companies to seek the protection of the law and be subject to taxes. A further such ‘legal fiction’ is extended to a ‘corporation sole’, in which case an office exercised by a person can have separate legal rights, duties and liabilities independent of who exercises that office; the president or prime minister of a country are the best examples.
Only an entity entitled to or subjected to personhood can be required to possess an ethic. Entities, living or otherwise, that have long been denied personhood by religion, society or the law, such as non-human living organisms, tools, mechanical devices, are not required to conduct themselves on ethical bases. This is dependent on agency, which is a related philosophical concept.
Agency is usually defined as the ability to take one’s own decisions and have the ability to understand their consequences. The agency of a human being who is a legal major, or a corporate entity, is taken to be fully formed. The agency of a growing child is considered to be developing, and they are therefore free from most liabilities and denied several rights until they attain the age of legal majority. The agency of an animal, plant, any other living organism, tool or mechanical device is considered not to exist. By law, they are property, either of individuals or corporate entities. If unclaimed, they exist as the property of the state by the principle of eminent domain.
You will realize that agency is independent of intelligence. The laws of most countries do not require a test of intelligence to confer agency (despite that being the basic assumption of why children are not treated on par with adults); hence the mentally challenged are not denied legal personhood upon reaching the age of majority. Thus, however much intelligence an animal or mechanical device might possess, it is attributed no agency, and hence no personhood.
This is the legal basis for determining the ‘ethics’ of an entity that possesses artificial intelligence (henceforth, AI). If the law (say a court or a parliament) confers personhood upon an AI device in the future (or as has already happened for Sophia in Saudi Arabia), it will immediately be entitled to the rights granted by the law, the responsibilities required by the state, and the liabilities imposed by the law. This is the only point at which ethical conduct is required of the ‘person’. In the absence of legal personhood, or even the limited personhood of childhood, any entity is liable to be treated as property. Remember, personhood is not enough for conferring legal liberty or certain rights. Until the enactment of the Thirteenth Amendment in USA, black slaves were person enough to bear liability, but not to have rights; similarly those imprisoned for crimes are usually held fully responsible for their actions even though their rights are severely curtailed.
The actions of something that lacks personhood altogether, and is thus a ‘chattel’ under law (allowing it to be bought and sold), will therefore be attributed to the ‘agency’ of its owner. This is the argument of the NRA when they say, ‘guns don’t kill people, people kill people’, however flawed you may find it. In such cases, the legal liability rests with the owner.
Why I went through all this discussion is to make this argument: however ethically trained an AI maybe, it will not be held liable until it is conferred personhood. So let us discuss two scenarios:
1. AI has personhood: In this case, the AI is responsible (and liable) for its actions. In this case, any ethical conundrum, it is expected to deal with. So if it is faced with the conundrum of killing an opposite, possibly innocent party to save the life of its ‘owner’ or employer, versus killing the owner/employer to save a large number of people, it is required to do what it thinks best and face the consequences in a court of law. In this case, all reviews and articles you have read should be checked for the assumption of personhood. This is also where the discussion of making instant calculations of the probabilities of the outcomes comes in. Remember, while human actions can be attributed to impulse or whim, an AI’s actions will always be calculated and deliberate.
2. AI does not have personhood at all: Then there is no case of ethics. An AI is a slave in this case, and whether owned by the state (say, to oversee electric supply or control traffic), or a private entity (like Facebook’s AI system) or an individual human (say as a driverless car), any actions of the AI will be held liable to its owner. It is thus but natural that such an AI be programmed to maximize the interests of its owner, even if it has to kill.
3. AI is conferred limited personhood: This would be a murky territory. Nevertheless, I think this is where Isaac Asimov’s Law of Robotics might fall, wherein robots have responsibilities, but no equality with humans. We guess (and we could be wrong), a lot of robo-ethics literature will be in this grey zone, even if the personhood is implied and not stated.
If we develop AI as specialized AI with no general intelligence, a lot of personhood debate might not be relevant and we would be dealing with other aspects of ethics of AI but we suspect that sooner or later we would have to deal with ethics of AI behavior to the point of personhood. Future holds immense promises and perils; needless to say there are exciting times ahead.
About:
Mr. Raamesh Gowri Raghavan is an award winning poet, a well-known advertising professional, historian, and a researcher exploring the interface of science and art. He is also championing a massive anti-depression and suicide prevention effort.
You can know more about Raamesh at:
https://sites.google.com/view/raameshgowriraghavan/home
https://www.linkedin.com/in/raameshgowriraghavan/?ppe=1
And here’s Raamesh telling his life story:
Raamesh and Sukant are working together on several projects on the intersection of science, technology, and art, and also projects on mental health.
Dr. Sukant Khurana runs an academic research lab and several tech companies. He is also a known artist, author, and speaker. You can learn more about Sukant at www.brainnart.com or www.dataisnotjustdata.com and if you wish to work on biomedical research, neuroscience, sustainable development, artificial intelligence or data science projects for public good, you can contact him at skgroup.iiserk@gmail.com or by reaching out to him on linkedin https://www.linkedin.com/in/sukant-khurana-755a2343/.
Here are two small documentaries on Sukant and a TEDx video on his citizen science effort.
This story is published in The Startup, Medium’s largest entrepreneurship publication followed by 277,446+ people.
Subscribe to receive our top stories here.
