Artificial intelligence (AI) research is moving incredibly fast and requires regulation. For instance, a recent study reversed a machine learning model that creates nontoxic drugs and made it create toxic drugs . This is called “dual use.” Other examples are unexplainable AI and hazardous incidents by autonomous cars or robot surgeons. These issues are why I chose to read this book. I wanted to be enlightened about the laws and regulations that should be applied to machine intelligence.
The fundamental problem treated in this book is the judgment-proof problem. In fact, “judgment proof” happens when a person’s belongings can’t be taken by the state when the person declares bankruptcy or fails to pay a debt. For instance, an exemption (that is, judgment proof) occurs if that person is unemployed or on a pension. The author translates the judgment-proof problem to intelligent machine owners and/or producers such that if the machine does harm, who can be blamed? What should the compensation be?
The book is composed of seven chapters divided into two parts. In the first part, the author introduces autonomous AI systems and why regulatory intervention is important. He discusses the nature, scope, and form of regulation and its limits. Also, in this conceptual part, the author introduces the fundamentals of his approach, which is based on economic reasoning and an economic analysis of law (comparative and behavioral law). The second part is the core of the book. In this part, the author tackles judgement-proof AI. In chapter 5, he speculates on the fears of a world ruled by AI agents and the possibility of AI taking over the world. For instance, he gives an example of a financial market dominated by autonomous trading algorithms. In chapters 6 and 7, the most important parts of the book, the author studies liability issues and offers a set of principles that can help mitigate risks and hazards and shape the law. He gives examples, shares historical inspirations, and provides recommendations for governments and policymakers so that the use of machine intelligence can remain both safe and efficient.
I like the style and format of this book: each chapter has an abstract, keywords, core material, and a bibliography. As a result, the reader can pick a single chapter, read it, and get a full grasp of what it adds to the subject. The book contains much literature and analysis. It is an excellent source of knowledge on regulating AI. It is an important read for lawmakers and computer scientists interested in ethical AI, as well as computer science historians who follow the evolution of this field.