AI 2041 combines fiction and nonfiction in a satisfyingly long and dense book that attempts to depict artificial intelligence (AI) two decades from now. The first component, by Chen Qiufan (Stanley Chan), consists of ten short stories in a genre termed “scientific fiction,” fusing fiction and popular science. After working at Baidu and Google, the author moved to writing sci-fi fulltime. His stories are interleaved with slightly technical afterwords written by Kai-Fu Lee, former head of China Google. The pandemic offered the authors a serendipitous opportunity to produce a collaborative book in record time and not have actual AI developments outstrip their prognostications.

I found the current state of AI outlined by Lee to be surprisingly robust, not shadowed by the three sellouts described in my review of Pelillo and Scantamburlo [1]. Lee does not shy from accounts of dangers, threats, and risks, but he also emphasizes the causes for hope. He is at his best when he champions creativity, empathy, and collaboration. In particular, I recommend the technical half of chapter 8.

Chapter 5 has an important self-reference to the book on pages 181 and 182:

[T]he AI writer was deliberately appealing to her taste with every turn of phrase. Flawlessness somehow made the stories far less exciting by depriving readers of challenges and surprises. To Aiko, these challenges and surprises were exactly what distinguished a good story from an ordinary one.

Chen skillfully manages the latter. Setting each story in a different country, Chen navigates cultures with ease and conviction, such as incorporating Fela Kuti, the pioneer of Afrobeat, into a story featuring deep fakes. The stories share a similar atmosphere, but the variation in characters and dilemmas forestalls boredom.

In view of multiple translators and editors, there is some surprising awkwardness in the book’s language. There are run-on sentences, acronyms before their embodiment, infelicitous word combinations (for example, “Primary enemy to the” on page 48). There are plot inconsistencies (In chapter 7, whose computer is doing what? In chapter 8, which girl is doing what?). Lee uses the word “externalities” several times. To me, an MIT math graduate who thereby escaped having to study Samuelson with Samuelson, the usage is vague. These are minor deficiencies, and overall the book is quite readable, entertaining, and elucidating--always avoiding technical jargon.

Some argue that the book should have included an evaluation of the use of AI, especially facial recognition (mentioned on page 26), in the surveillance of China’s population and in particular of Uyghurs. More details are given in Olcott [2]. Kahn [3] includes interviews with the authors, and this topic was not mentioned. Chen assessed the dark and disquieting aspects of his stories as warmer than his previous work: “Compared to other stories I wrote before, this is the brightest thing I have ever written.” Lee defended his previous book, AI superpowers [4], as an analysis of business competition, not US-China political relations. On the current book, he suggested to Kahn that Americans might be more disturbed by the characters’ losses of autonomy and privacy than Asians, who tend to have more collectivist cultures according to Lee.

Kao [5] interviewed Lee, “who referred to the secret sauce of humans in the AI era as ‘warm skills.’ ... These include empathy, compassion, collaboration, a growth mindset, agility, trust building, and creativity among others--what some refer to as ‘21st century skills.’”

Lee’s signal attitude is optimism. For example, he states on page 118:

What’s important is that we develop useful applications suitable for AI and seek to find human-AI symbiosis, rather than obsess about whether or when deep learning AI will become AGI [artificial general intelligence, like unto human]. I consider the obsession with AGI to be a narcissistic human tendency to view ourselves as the gold standard.

On the same page, Lee emphasizes the monumental impact AI will have on education, ending with the assertion that “this symbiotic and flexible new education model can dramatically improve accessibility of education and also help every student realize his or her potential in the age of AI.” They assert other dramatic changes will occur in work, entertainment, mentation, and relationships.

In the final chapter’s discussion of plenitude, the phenomenal amount of discarded food mentioned on page 429 reminds me of White Pony Express (https://www.whiteponyexpress.org), an organization that has rescued 20.4 million pounds of food in eight years.

I advise the reader to ignore a short book called Summary of AI 2041.

More reviews about this item: Amazon