2022.Jan.28

REPORTS

Report on the 5th B’AI Book Club
Meredith Broussard, Artificial Unintelligence: How Computers Misunderstand the World (2018)

On October 26, 2021, B’AI Book Club, a book review session by the members of the B’AI Global Forum, held its fifth meeting online. Kayoung Kim, Project Researcher of the B’AI Global Forum, introduced Meredith Broussard’s book Artificial Unintelligence: How Computers Misunderstand the World (The MIT Press, 2018).

This book can be positioned as a counter argument against technochauvinism, which is a belief that technology is always the best solution even to the complicated social issues, and a tech culture that is strongly rooted in the belief, as Artificial Intelligence permeates every corner of our lives today.

Based on the knowledge and insights she has gained through practical works in the field, the author, a computer scientist and data journalist, argues that we should always bear in mind that “the objectivity and fairness of algorithms are myths and that there are limits to what technology can do.” As the basis of this argument, she presents her own findings which revealed structural reasons why poor schools cannot perform well at standardized tests, insists that such problems cannot be solved by AI, and explains in detail how machine learning works to indicate that machines are not actually perceptive learners so that a lot of pitfalls can exist in predictions made through mere mathematical calculations. Furthermore, she criticizes the self-driving car, which has been considered as an example of AI technology for a safer future society, saying that it is completely incapable of dealing with situations where ethical decisions are required.

However, the author emphasizes that the reason for trying to understand what technology cannot do is not to be pessimistic about the age of AI, but rather because doing so will give us a clearer picture of what it can do. In the latter part of the book, she cites computational journalism as an example of an area in which human-computer collaboration can yield meaningful results, describing a project that built a software program to analyze and visualize the flow of campaign funds during the 2016 US presidential election, which provided journalists with useful data. The book thus presents the most appropriate attitudes toward technology that should be adopted for a better future.

After the presentation, participants mainly discussed how this book could be used in the context of B’AI Global Forum’s activities. While the book was highly evaluated for its writing style and structure, which made it easy to understand for those who had no knowledge of machine learning or algorithms, and for the abundance of concrete examples, there was an opinion that it was not suitable for use as a textbook in graduate school for exactly the same reason. Also, it was pointed out that it is not clear whether some frequently used words such as “technochauvinism” and “General AI / Narrow AI” are the author’s own expressions or technical terms with academic origins, which could be one of the reasons why it is difficult to use this book as a textbook.

However, the author’s background as both a computer scientist and a journalist makes her argument highly convincing. She argues about the use of AI in journalism not from a business perspective but from one of its contribution to society, which overlaps with the mission of the B’AI Global Forum. In this sense it can be said that this book is undoubtedly a thought-provoking.