1. AI is not something which makes computer bigger than or beyond the control of mankind. Its created and still controlled by humans only. The term ” Machine Learning” is a new epithet for what is conventionally called ” self evolving”. By feeding more and more data, information and logical functions into a machine it can be made to become self evolving OR what the new age generation says “Machine learning”.
2. Because AI will not become a master to win over the universe (mankind) it still stays in some sort of Master slave configuration Or so to say Principle-Client relationship. Very clearly in such a case the indirect or Vicarious liability for the client will lie on the principle or that of a slave on the master. No new legislation is envisaged for such a creation.
3. Now coming to the term “Legal person” as the Robot Sophia in Saudi has been christened and has been given citizenship. A duly incorporated company is also a legal person or a juristic person having its own entity, rights and liabilities. And we have Laws for that in Company law and various other Acts. The status of A machine under AI and a incorporated company is same both being Legal person. Just as the company law puts liability on Directors of the company for various liabilities of the company similarly the owner(s) of a machine/system with AI can be made responsible/accountable/liable. Its wrong to construe AI as a fired missile no longer in control of the firing hands. No absolutely not. In your example of auto driven car on AI if it commits some accident then surely the liability should lie with the owning company or the individual owning it. After all we are not producing AI machines to have a place of their own in the society leaving them to become their own masters or sovereign machine men. They will at best be juristic persons not humans.
4. A more dangerous and concerning issue is of producing designer babies through genetic engineering. AI is nothing before it. Genetic engineering and causing to make tailor made babies will ACTUALLY need appropriate legal handling. That is the most worrying technological development that needs Jurisprudence and yet the world is not able to find a proper legal framework to look at it. AI is a small fry in the pan before this issue.
AI, in my opinion, does not need a separate Law since it is not something beyond the control of its master (human). As in case of other legal persons (such as a company or a robot) AI can be seen from that angle and can be properly handled in the existing legal framework and jurisprudence.