With terms like artificial super intelligence, brain-computer interface, and cyborg models peppered throughout, Nell Watson’s book Taming the Machine: Ethically Harness the Power of AI may make readers feel like (or wish) they had wandered into the sci-fi section of their local bookstore. But these terms along with others like robotic avatar, or my personal favorite— stochastic parrot—aren’t the creation of Isaac Asimov but instead are part of the fascinating, terrifying, and ever-changing world of AI.
Taming the Machine provides two things desperately needed in an increasingly AI-driven world: hope and information. Never does Watson diminish the potential AI has to catastrophically change the world, likening it to the advent of airplane travel in the 1950s: “Similar to air travel in the 1950s, Generative AI is an exciting new technology with a high likelihood of tragedy.” That said, the history of air travel has already been written; the history of AI is arguably still in progress. While the likelihood of tragedy may be high, it is not written in stone. So Watson promises, “With care and knowledge, avoiding either complacent optimism or defeatist pessimism, we can effectively harness AI’s potential.”
Thus begins Taming the Machine, which opens with chapters on the history, the promise, and the peril of AI. While it might be tempting to flip to the end, after all a chapter titled “Sympathy for the Machine” is hard to resist, these chapters are like building blocks, each supporting the next.
Chapters are detailed and fully researched with multiple examples and definitions; in fact, Watson includes so much information, just deciding where to start this review was a bit daunting. Of course, that’s AI—it’s everywhere, and it seems to change every day. Perhaps this is why Watson describes machine learning as a “modern-day oracle” and AI as “a game-changing force with the ability to see into the future.” AI, Watson notes, can make life and death medical decisions, make five-day weather forecasts as accurate as the one-day forecast, predict economic changes, and help us understand the language of bats and chickens (among many, many other things). AI even appreciates good grammar; according to Watson, AI prompts should use active voice and follow parallel structure.
At first glance (and this is a book that deserves multiple reads and most likely a permanent spot on the bookcase), the target audience appears to be corporate and particularly corporate leadership. For example, the chapter titled “Preserving Privacy” reminds organizations of the importance of getting clear consent from people before using or gathering their data, and ends, like all chapters do, with easy-to-read text boxes that break summaries and action items into three categories: “The bottom line,” “The big picture,” and “Leadership action points.”
That said, a closer examination shows plenty for a more general audience. Anyone who values privacy might appreciate knowing that personalized experiences can go far beyond Netflix recommending a television series or a movie. Watson’s example: the possibility a company could use someone’s online searches for breakup songs as a reason to flood their feed with alcohol advertisements.
The book could also help people—particularly employees—be better apprised of their own rights. Watson is often blunt, usually refreshingly so, in her statements to big business. In the chapter “Fairness and Franchise,” Watson states, “Employers should be careful to avoid strict social media policies which infringe on employees’ autonomy and freedom of speech. Employers need to strike a balance which reasonably protects their interests without unduly limiting their employees’ rights to participate in public discourse, especially in digital spaces where diverse norms and viewpoints coexist.” Good information not only for employers but also for employees who may feel pressure to censor their thoughts online.
Balance is something Watson clearly believes is necessary to harness AI’s potential. Another familiar theme in the book is that AI is a double-edged sword. In one of the first references to the double-edged nature of AI, Watson states “On one side, [AI] offers unprecedented advantages in data handling, problem-solving and even pushing the boundaries of human creativity. On the other, there are very real limitations, such as AI’s tendency for confabulation, which could have serious repercussions.” AI is not the only double-edged sword in this book, however. Regulating AI is also likened to a double-edged sword: “On one hand, [regulation] can enforce best practices and ensure safety; on the other, it could stifle innovation and favour established companies at the expense of smaller players.”
If concerns about favoritism, confabulation, AI taking our jobs, or companies knowing our intimate medical issues aren’t enough, this book shows that there is so much more to worry about, such as new forms of discrimination. Considering the amount of discrimination in real life, it probably shouldn’t be surprising that we have algorithmic discrimination as well. Algorithms can be racist, sexist, ageist, or as Watson specifically discusses, show bias based on someone’s political affiliations.
Still as the saying goes, knowledge is power, and for every (or almost every) problem, Watson proposes alternatives, solutions, or new approaches. She argues for transparency, ethics, trustworthiness, and human governance. Her writing style is often both clever and personal and usually, given the subject matter, unbelievably clear. She invites us on this journey with her and often her comparisons, likening AI to everything from salt to unpaid interns to magicians who never reveal their tricks, are not only spot on but are often creative. Illustrations, such as the one below, add some visual interest as well.
Watson looks at the past and present but also turns to the future—comparing the evolution of AI to Darwin’s theory of evolution and wondering if eventually AI will make us rethink our place in the cosmos. Finally, never does Watson suggest more technology is the only solution—one of her last action items is to “[r]ecognize the wisdom encoded in many ancient practices. Engaging in regular introspective practices like meditation, journaling or prayer will help clarify the murky waters of your inner voices.”
If your inner voices are still murky, or if AI still has you slightly terrified, you might find it hard to believe the chapter on humanizing machines, which suggests AI eventually could be as warm and loyal as a golden retriever.
In that case, it might be helpful to recall some of Watson’s reassuring words from the introduction:
“Humanity has, in the past, negotiated nuclear test ban treaties, ozone layer and acid rain treaties, and dampened age-old conflicts, such as Northern Ireland. Perhaps we can obtain similar wins with the governance of responsible AI.”
This book can be purchased on Amazon, or directly from KoganPage.
After graduating from Auburn University, Catherine Ramsdell became an educator. Her interests are oddly varied, but she enjoys teaching anything that involves a good story—from brand storytelling to mythology to data journalism.
