The Atomic Human

July 1, 2024
A respected expert in the field argues that AI systems are powerful tools, but it takes a human mind to deal with life's complexity. Debates over AI often devolve into advocates insisting that it is the cure for all social ills and opponents arguing that the technology constitutes an existential threat to society. Lawrence avoids the either/or paradigm, and he certainly has the experience to paint a more nuanced picture: He is currently a professor of machine learning at Cambridge, and he previously served as the director of machine learning at Amazon. The author's chief concern is that AI systems could eventually take over human decision-making, although ultimately they are in the hands of massive companies seeking profit. However, AI is inherently unable to collate crucial information with sufficient subtlety, vision, and even intuition. One of the author's comparisons is Eisenhower's decision to launch the D-Day invasion: "Eisenhower didn't have complete information, he only had 'the best information available.' His decision required judgement: at the time he made it he knew he could be wrong, and being wrong would have dreadful consequences for thousands of soldiers and the long-term course of the war. Judgements of this form remained firmly the preserve of the human." Lawrence advances a host of important arguments, but he repeatedly drifts away from his theme. The text sometimes feels like a stew of cultural references, as the author discusses novels, poems, and historical events that have little discernible connection to AI issues. The digressions and references make the narrative difficult to follow. The book would have been much improved by a strong editorial hand to keep the author on mission and create a shorter, more focused work. Lawrence is a major figure in the AI landscape, but his book is marred by a lack of discipline and narrative organization.

COPYRIGHT(2024) Kirkus Reviews, ALL RIGHTS RESERVED.

July 15, 2024
Lawrence, a machine learning professor at Cambridge University, debuts with a muddled inquiry into what distinguishes human decision-making from artificial intelligence. Relying on convoluted analogies, Lawrence details at length the uncertain intel and logistical complexities Dwight Eisenhower had to contend with when deciding when to invade Normandy, but barely develops his comparison between Eisenhower and AI, rendering his assertion that the technology lacks Eisenhower’s capacity for “judgement” somewhat nebulous. A particularly baffling chapter explains how participants in a 2006 clinical drug trial suffered debilitating side effects to make the simple point that, just as the participants should’ve been less trusting of the doctors conducting the trial, humans should be wary of AI algorithms they don’t understand. Lawrence embarks on a dizzying array of tangents touching on the invention of calculus, the fly-ball mechanism that automated steam engines, behavioral psychologist Daniel Kahneman’s cognitive theories, and the European squid’s propulsion mechanics, but their relevance to AI is often oblique at best. Lawrence also equivocates so much (he contends he’s “an optimist about AI” even as he decries tech companies for “deploying software systems that cannot be controlled”) that it’s hard to tell where he stands. This gets lost in the weeds. Agent: Max Edwards, Aevitas Creative Management.