The topic of this book is artificial intelligence (AI), one of the most rapidly developing fields in the last decade. The book was written out of a belief that the rise of computing and AI is of epochal significance. AI proves, through successful applications, its potential for rapid economic, social, and cultural development in many areas, including healthcare and energy sustainability. These advancements, however, have also brought about some concerns related to possible harm, both material (such as safety, health of individuals, property damage) and immaterial (loss of privacy, limitations to some human rights, discrimination).
Brian Cantwell Smith is a professor known for his research on the foundations and philosophy of computing. His book is a true reflection of this focus. Smith attempts to provoke thoughts on AI. He aims for a deeper understanding of AI than we have today, addressing issues such as what AI is, what intelligence requires, and so on. This book can be viewed as an attempt to lay out some understanding of AI, which is a prerequisite to meaningful conversations about the ethics of AI according to Smith.
The narrative of the book is interwoven with two concepts: the formidable reckoning achieved by machines, and judgement attributed currently to humans. Smith discusses automated reckoning systems and automated judgement. He does not believe that automated judgment is on any immediate horizon. Readers can follow his arguments for this belief.
The book contains 13 chapters, a list of references, and an index. Readers will also find a wealth of additional sources and comments.
Smith examines the history of AI from its first-wave origins (“Good Old Fashioned AI” or GOFAI) to machine learning and today’s approaches. He first provides background, including methodological preliminaries and questions about the future (chapter 1). Chapter 2 introduces symbolic representation, while chapter 3 discusses its failure. The main topic of chapter 4 is a transition from symbolic AI to subsymbolic AI. Chapter 5 looks at machine learning compared to GOFAI. The next chapters discuss particular concepts related to AI development: assessment (chapter 6), epistemological challenges (chapter 7), objects (chapter 8), and the world (chapter 9).
Chapter 10 explains the book title’s “reckoning and judgement.” It is followed by chapter 11’s discussion of seven comments: human versus machine, details, consciousness, emotion, responsibility, ethics, and construction. Next, chapter 12 applies these seven comments to three issues: reference, context, and ground of being--the world. Chapter 13 concludes the book with some thoughts on the learning path through AI development. Smith argues throughout the book that we should use AI for the reckoning tasks at which AI systems excel, while also strengthening our commitment to judgement, ethics, and the world.
I like Smith’s calls to which he hopes the book will draw attention: “(i) that we should learn how to use AI systems to shoulder the reckoning tasks at which they excel, and not for other tasks beyond their capacity, and (ii) that we should strengthen, rather than weaken, our commitment to judgement, dispassion, ethics, and the world.”
Overall, this is a great book for readers interested in a holistic view of the rapidly developing field of AI. It develops some intellectual tools for assessing current developments in AI. It is more philosophical than technical. On the other hand, Smith explains some technical concepts in the context of computing and human and intelligent behavior. I recommend the book to all thinkers, especially students of technical and non-technical disciplines who are interested in a deeper understanding of machines and automatic reckoning processes in relation to judgement.
