On October 25, 2017, Sophia became the first robot in history to be a full citizen of Saudi Arabia. First of all, let’s assume there is actual citizenship granted, and it is not just a marketing stunt. And so we look at what happens if we create such legal rights in robots. The first problem I see with this is we like to think about robots as being humanoid. The thing humans do most and do best is the understanding of the humans. The largest part of our brain is devoted to that. As a result, it means then is that when we imagine artificial intelligence, the first thing we do is imagine humanoid robots like Sophia. We imagine terminator, C-3PO, we imagine Skynet, hell 9000, all humanoid types of robots. More realistically: robots aren’t humanoids.
LEGAL RIGHTS IN ROBOTS: PROBLEMS AND IMPLICATIONS
So just even looking at Sophia: she has some really cool facial abilities, she can move her mouth, she is humanoid, she can move around. She is a fantastic robot. But that robot is kind of like a puppet. She can move around, but then outside of Sophia, there is a chatbot server, probably hosted on the cloud, having some onboard computers. Obviously, there are human creators that amend Sophia from times to times. There might be various systems that change. Now, as someone who develops chatbots such as Ailira, I am familiar with it, and there is some fantastic technology, but we are still a long way from having a chatbot that is a human. For example, a simple conversation:
“Hi”
“Hi”
“How are you?”
“Great”
“That’s good”
While I get one response in and I get another response out, there is some really cool math that goes into understanding natural language, and we can see some fantastically developed chatbots. But still, a chatbot doesn’t have any kind of consciousness or thinking. It’s purely a program for interacting. And in fact, you know they are often built on deception when you ask questions like: “What’s the sky like today?”. They can’t answer. Or answer: “What do you think it’s like?” To develop a really neat chatbot at the end of the day, that’s not a human.
They’ve got a really cool onboard computer as well. The question here is: what have you granted a citizenship to? Is it a puppet that is Sophia’s body? Is it their onboard computer that controls that? Does she have the chatbots server in the cloud? Let’s just think through the implications. For example, if we replace Sophia’s body. Do the replacements hold the citizenship, or is it just her? Is it the computer that controls it, that interfaces it? Or is it the cloud? What happens if I take this cloud server and I copy it? Which one is the original Sophia?
PRECEDENTS OF ARTIFICIAL LEGAL PERSONHOOD
Now, I don’t have a problem with granting citizenship or legal personhood to something artificial. There is a long precedent in that, and we see that today in companies. Companies have an artificial legal personhood, they can own assets, they can drive income, they get taxed. Ultimately, there are shareholders and directors behind them. Finally, it comes to a human somewhere. But we might have a company that owns a large amount of property that last for hundreds of years. They can go bankrupt, and they can “die.” We can answer questions of how we copy the company, but the thing about the company – it is clearly defined. There is a register or some kind of other legislation. And that registration defines what is the company. Furthermore, I can copy the company’s constitution, its rules, its “programming,” as we might say, but I would have to register it again.
Coming back to the concept of what we do with a robot. How could we register a robot as a separate person? “Citizen” is a very strong word, but what happens if you could give a robot a legal personality like a company? So, we could give it to the server, in which case you would have to take some kind of programming. In addition, this registered program is itself a separate entity, and it can drive income. It can sue or hold assets or has rights and obligations. That is very different to the idea if we have a humanoid robot that sits in itself and that person is a citizen.
Let’s have a complex scenario here. What happens if we have more industrial robots, and they are all controlled by one centrum server. So here, what would you give a robot personhood to? Each arm? It could be hundreds; it could be thousands. Would it make sense to give it to the server, the program? We like to think of physical things, and that’s why I think Sophia is so attractive and fantastic story, but what we need to come back to if we were gonna give some kind legal personality, I would suggest it would be the “AI” itself. Consequently, it would need some kind of registration. Because otherwise, if we cloned it, I could suddenly create 5 million different AIs and imagine each of them earned only one dollar. I could suddenly earn millions of dollars. And pay only a tiny amount of tax on a progressive tax rate.
On the other hand, the other problem is giving a robot legal personhood. What happens if Sophia creates her own Sophia? Sophia 2. What happens if she goes out and makes an exact replica of herself, connects it up to that servers? Is this Sophia 2 a citizen as well? She has legal personhood? What happens if I attack Sophia? If I damage her? Is that assault? Can I marry her?
SEPARATE LEGAL STATUS FOR AI
These questions can be answered if you look at a company. A company can hold assets; it can sue itself, so we can solve these things by looking to company law and artificial entity. We can register it somehow, and we can take this entity and say: this is a separate legal person, but note that whatever we are doing, we are standing it on some imaginary thing. Maybe it’s the server; maybe we have a registered program. Hence, I think there is quite a good possibility as we get a more advanced AI that we will create a separate legal status for this. But, I think we should always remember not to get caught up in a hype of a humanoid-looking robot.