Book Notes: Co-Intelligence: Living and Working with AI

Note: This brief quasi-book review/book synopsis is the first item in an experimental new section of SSTI’s newsletter, potentially joining other regular sections such as Useful Stats, Fed/Leg News, State News, Member Updates, and Recent Research. Its periodic continuation after the contributions we present over the summer will depend on feedback from our members and Digest readers. Comments may be shared with skinner @ ssti.org

Ethan Mollick, a Wharton professor specializing in entrepreneurship and innovation, knows people are thinking a lot about generative artificial intelligence (AI). He recognizes many are worried, as his introductory chapter “Three Sleepless Nights” of his book, Co-Intelligence: Living and Working with AI, reveals. Mollick’s opening sentence tells readers those 72 hours are the minimum price he thinks anyone will pay for “really getting to know AI.”

Yet another doom and gloom book about the everything-changing platform technology? Not at all.

Co-Intelligence presents in accessible language clear looks from both sides—all sides?—of what AI is, what it is doing and not doing to society, and what we can do with and about it. The book is full of warnings but also contains plenty of pragmatism about the inevitability of AI continuing to evolve, improve, encroach, and impact modern living. In fact, the fourth of his four principles of co-intelligence (the human-AI interface or relationship) is “Assume this AI [the version you’re currently using at any given time from now on] is the worst AI you will ever use.”

The short book, published in April 2024 and logging 256 pages from cover to cover, is divided into two sections. Part One introduces AI as an Alien, not alien as a metaphor for AI, but really as an alien mind that we need to learn to live and work with. He points out it could be scary, “It’s important to note that the process is not without human cost.”

Mollick’s main point, however, is that we must continue to interact with AI (principle #1), be involved in its development and evolution (principle #2), and respect it while striving to guide its maturation as a good parent would its child (principle #3). We must teach and reinforce ethics, integrity, honesty, and fair play. That might be a tall order in a market economy that increasingly measures a human’s value by their monetary value. The author points out there is no reason to think AI won’t change its ethics and character as it matures to need humans less and less, other than to pay the electric bill to keep it running.

Part Two comprises the majority of the book, describing the alien that is AI through five roles: as a person, a creative, a coworker, a tutor and a coach. Useful examples of generative AI in action to support Mollick’s points—both pro and con—are presented in each chapter.

The closing chapter, AI as Our Future, explores four different scenarios that the author can envision as co-intelligence unfolds. I won’t say anything specifically about any of them because, as he says in the epilogue, it’s up to humans and how we choose to live with the aliens we’ve created.

In addition to print and digital editions, Co-Intelligence is available as an audiobook read by the author.

In my opinion, the straightforward explanation of how generative AI is evolving, examples of it at work, and a good balance of the scary and beneficial aspects of the alien makes Co-Intelligence a worthy exploration of AI for the TBED practitioner sorting out the policy implications of AI for the field. Sleepless nights, indeed.