What do you think?
Rate this book
The Reasonable Robot: Artificial Intelligence and the Law
Today, artificial intelligence (AI) and people do not compete on a level playing field. From a safety perspective, AI may be the best choice for driving a vehicle, but laws often prohibit driverless vehicles. At the same time, a person may be better at packing boxes at a warehouse, but a business may automate because AI receives preferential tax treatment. Or, AI may be better at helping businesses to innovate, but these same businesses may not want to use AI if doing so restricts future intellectual property rights. In The Reasonable Robot, Ryan Abbott argues that the law should not discriminate between people and AI when they are performing the same tasks, a legal standard that will help to eliminate market distortions and to ensure that decisions are made on the basis of efficiency. This work should be read by anyone interested in the rapidly evolving relationship between AI and the law.
Unknown Binding
Published June 17, 2020
16 people are currently reading
89 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 6 of 6 reviews
April 23, 2022
I heard Mr. Abbott speak a couple of weeks ago and felt that I should get a more in-depth understanding of his thinking by reading his book. The talk that I heard was focused on intellectual property law, but in the book he has a broader range, also looking in detail at the implications of AI for tort law and tax law.
On one level, Mr. Abbott's analysis is interesting and smart. Wouldn't it make more sense to look at ways for the law to provide building blocks that would allow society to self organize to create the best outcome for everyone? And in this regard, wouldn't it be better to adjust tax law to be neutral between humans and AI so that it doesn't favor automation as it presently does? In IP law, if allowing AI to be a legal author or inventor would give us more creative works of authorship and socially useful inventions then shouldn't we put aside angels-dancing-on-the-head-of-a-pin debates about whether a non-human can hold IP rights? And in a world where a machine has a form of intelligence and can in many cases make better decisions than humans, doesn't it make sense to hold humans to the standard of care that they could have achieved if they used a machine and to hold the machines to a standard of care that reflects what a reasonably conceived and implemented state of the art machine would do? Mr. Abbott has thoughtful answers to all of these questions.
But on another level, I couldn't help thinking that Mr. Abbott was barking up the wrong tree. The three areas of the law that he focuses on - IP, tax and torts - are three of the most broken areas of our legal system today. IP law and tax law are structured in ways that favor big corporate interests and wealthy people; tort law allows for massively time-wasting lawsuits and extortionate settlements that don't seem to do a lot to improve behavior or create social balance. And Mr. Abbott is suggesting ways to put band-aids on these broken systems that won't make them any less broken in the new era of AI. In fact, one of the biggest problems with all of these areas of the law is that they are gamed by people with enough resources to figure out ways to twist them to advantage. AI is going to make gaming the system even easier. And Mr. Abbott relies on the kind of legal thinking that I was taught in law school that seems very reasonable on the page but that may or may not have much of a relationship to behavior in the real world or to establishing a justice system that dispenses something that resembles justice.
I'd love to see a sequel book that takes a more radical approach to addressing the issues that Mr. Abbott discusses and that considers ways that AI could be used to create a very different legal structure that is built on basic principles that would reduce the abuses of the present system.
On one level, Mr. Abbott's analysis is interesting and smart. Wouldn't it make more sense to look at ways for the law to provide building blocks that would allow society to self organize to create the best outcome for everyone? And in this regard, wouldn't it be better to adjust tax law to be neutral between humans and AI so that it doesn't favor automation as it presently does? In IP law, if allowing AI to be a legal author or inventor would give us more creative works of authorship and socially useful inventions then shouldn't we put aside angels-dancing-on-the-head-of-a-pin debates about whether a non-human can hold IP rights? And in a world where a machine has a form of intelligence and can in many cases make better decisions than humans, doesn't it make sense to hold humans to the standard of care that they could have achieved if they used a machine and to hold the machines to a standard of care that reflects what a reasonably conceived and implemented state of the art machine would do? Mr. Abbott has thoughtful answers to all of these questions.
But on another level, I couldn't help thinking that Mr. Abbott was barking up the wrong tree. The three areas of the law that he focuses on - IP, tax and torts - are three of the most broken areas of our legal system today. IP law and tax law are structured in ways that favor big corporate interests and wealthy people; tort law allows for massively time-wasting lawsuits and extortionate settlements that don't seem to do a lot to improve behavior or create social balance. And Mr. Abbott is suggesting ways to put band-aids on these broken systems that won't make them any less broken in the new era of AI. In fact, one of the biggest problems with all of these areas of the law is that they are gamed by people with enough resources to figure out ways to twist them to advantage. AI is going to make gaming the system even easier. And Mr. Abbott relies on the kind of legal thinking that I was taught in law school that seems very reasonable on the page but that may or may not have much of a relationship to behavior in the real world or to establishing a justice system that dispenses something that resembles justice.
I'd love to see a sequel book that takes a more radical approach to addressing the issues that Mr. Abbott discusses and that considers ways that AI could be used to create a very different legal structure that is built on basic principles that would reduce the abuses of the present system.
Read
November 3, 2021The Reasonable Robot is a fascinating look into the legal challenges posed by artificial intelligence (AI) and how the law may transform to address these challenges. Over the past few years, the use of AI has become increasingly prevalent. This is in part due to “more advanced software, greater computing power, and growth of big data” (p. 30). However, the law surrounding AI has not changed to reflect this. The central argument of the book is the need for “a new guiding tenet to AI regulation, a principle of AI legal neutrality” (p. 3) that will “help the law better achieve its underlying goals” (p. 134).
Despite the absence of a general, well-accepted definition of AI, a definition is required to ground the exploration of how AI should be regulated. For this book, Abbott defines AI as “an algorithm or machine capable of completing tasks that would otherwise require cognition” (p. 22). Abbott stresses throughout the book that how the laws surrounding AI should change depends on the relative sophistication of the AI being used.
The book begins with an exploration of legal personhood for AI. Abbott does not advocate for AIs having legal rights or personhood unless such would benefit people. Following this, Abbott looks at four areas where the law in its current state fails to balance the needs of humans and users of AI. These areas are tax, tort, intellectual property, and criminal law. Key issues in these areas include liability and legal personhood, and Abbott makes several suggestions as to how the law needs to transform to make the adoption of AI in these areas “fair.”
Abbott notes that the law distinguishes “often inadvertently, between behaviour by people and AI and ... this has unintended consequences” (p. 134). These unintended consequences arise from the position that AI and humans are not the same and that they have different motivations and abilities. This is given the conjecture that motivation may even be attributed to AI. This lack of motivation or intent complicates the process of assessing the apportionment of liability and, consequently, the application of criminal and tort law.
Abbott argues that applying a strict liability standard for AI, rather than engaging in the process of determining whether the AI’s conduct was unreasonable, makes sense as this standard would not require proof of intent. To assess liability, Abbott argues that it is more ideal to focus on the outcome of an AI’s actions rather than on its design. This is because of the innate difficulties to be faced in assessing whether the AI’s design is flawed, such as acquiring access to the AI proprietary software for analysis.
The book concludes with a discussion of bias in AI. AI may inherit human biases either consciously, such as through a deliberate decision by the programmer, or unconsciously, such as through using biased historical data. In theory, AI should be less biased than a human. As Abbott notes, “[t]he human mind is, perhaps, more [of an] algorithmic black box than AI” (p. 138). However, given that it can be challenging to determine how a sophisticated AI arrived at a decision, it may not be possible to figure out whether an AI is intrinsically biased.
Although Abbott is a law professor at the University of Surrey, the book is written for the layperson with only the occasional lapse into scholarly or academic language. The book is written from an American point of view, with the examples Abbott uses spanning U.S. tax law, U.S. product liability law, and other areas. Although the book is written with the U.S. legal system in mind, it is a thought-provoking read no matter what jurisdiction one is from.
Despite the absence of a general, well-accepted definition of AI, a definition is required to ground the exploration of how AI should be regulated. For this book, Abbott defines AI as “an algorithm or machine capable of completing tasks that would otherwise require cognition” (p. 22). Abbott stresses throughout the book that how the laws surrounding AI should change depends on the relative sophistication of the AI being used.
The book begins with an exploration of legal personhood for AI. Abbott does not advocate for AIs having legal rights or personhood unless such would benefit people. Following this, Abbott looks at four areas where the law in its current state fails to balance the needs of humans and users of AI. These areas are tax, tort, intellectual property, and criminal law. Key issues in these areas include liability and legal personhood, and Abbott makes several suggestions as to how the law needs to transform to make the adoption of AI in these areas “fair.”
Abbott notes that the law distinguishes “often inadvertently, between behaviour by people and AI and ... this has unintended consequences” (p. 134). These unintended consequences arise from the position that AI and humans are not the same and that they have different motivations and abilities. This is given the conjecture that motivation may even be attributed to AI. This lack of motivation or intent complicates the process of assessing the apportionment of liability and, consequently, the application of criminal and tort law.
Abbott argues that applying a strict liability standard for AI, rather than engaging in the process of determining whether the AI’s conduct was unreasonable, makes sense as this standard would not require proof of intent. To assess liability, Abbott argues that it is more ideal to focus on the outcome of an AI’s actions rather than on its design. This is because of the innate difficulties to be faced in assessing whether the AI’s design is flawed, such as acquiring access to the AI proprietary software for analysis.
The book concludes with a discussion of bias in AI. AI may inherit human biases either consciously, such as through a deliberate decision by the programmer, or unconsciously, such as through using biased historical data. In theory, AI should be less biased than a human. As Abbott notes, “[t]he human mind is, perhaps, more [of an] algorithmic black box than AI” (p. 138). However, given that it can be challenging to determine how a sophisticated AI arrived at a decision, it may not be possible to figure out whether an AI is intrinsically biased.
Although Abbott is a law professor at the University of Surrey, the book is written for the layperson with only the occasional lapse into scholarly or academic language. The book is written from an American point of view, with the examples Abbott uses spanning U.S. tax law, U.S. product liability law, and other areas. Although the book is written with the U.S. legal system in mind, it is a thought-provoking read no matter what jurisdiction one is from.
August 10, 2021
I heard this book being discussed on a podcast and became unexpectedly fascinated about the question of how IP law reacts to increased AI sophistication. So I downloaded the book on kindle. It’s an involved read, and by the end I was losing concentration and keen to move onto a more fun evening book. However the book raises lots of interesting aspects of AI that I hadn’t considered before and the IP chapters in particular were very thought provoking. Definitely a book that requires some mulling over.
September 21, 2020
I've been in tech for many decades, and have followed AI for most of it. This book is a lawyer's view of how the legal environment should change in the wake of AI. While I disagree with a number of the author's conclusions, he does a great job of laying out the key challenges involved in the areas of taxation, liability and invention. I don't have to agree with him in order to see this book as a good introduction to the issues and something to help start the debate.
October 8, 2020
This is a well-researched, comprehensive and interesting book about the relationship between AI and the law, but it requires close reading. I hope to write a longer review when I have more time next week.
I received this free ebook from NetGalley in return for an honest review.
I received this free ebook from NetGalley in return for an honest review.
June 8, 2023
Informative and interesting. brings up great points about the legal implications of AI and the ways in which the law ought to be examined in order to build a more equitable world that both embraces AI and prioritizes human wellbeing.
Displaying 1 - 6 of 6 reviews