The AI con How to fight big tech's hype and create the future we want

There are basically two ways of thinking about artificial intelligence. AI will either catapult humanity into new realms of productivity and effectiveness or hasten society's extinction. With respective backgrounds in linguistics and sociology, Bender and Hanna tackle this dichotomy from the standpoint of ethics and education. In short, breezy chapters, they examine AI from basic principles to bedrock definitions, expose marketing hype, offer healthy doses of reality, and present grounded methods for accepting the limitations and successful uses of AI. The autofill function on text applications and the chat pop-ups offering aid during an online shopping spree are two common examples of LLMs, large language models, that have made AI more palatable. But for every fuzzy widget, there are myriad mechanisms governments and corporations can use to integrate AI in ways that can seize control of content and override privacy. With the creation of new governmental policies and agencies seeking efficiency efforts heavily reliant on AI, this well-reasoned, accessible guidebook is a foundational resource for understanding AI now and on the horizon.

The bots probably aren't going to kill you, but they're probably not going to save you, either. Linguist Bender and sociologist Hanna, the founders of a merrily debunking podcast about all things AI, write that there's hoopla aplenty about how AI will make our lives better--or perhaps worse. In an opening salvo, they write, gamely, "each time we think we've reached peak AI hype---the summit of bullshit mountain---we discover there's worse to come." There are real concerns, of course, especially for people of color, whom AI facial recognition algorithms are altogether too likely to identify as crime suspects and who are likely to be judged risky candidates for pretrial release if they've been charged. Those "daily harms being done in its name" are more profound than a feared robot apocalypse, as are other sequelae of AI: the replacement of human workers with machines, the shredding of career tracks with gig work, the collapse of creative industries. (Actually, the authors add, AI probably won't replace your job, "but itwill make your job a lot shittier.") Those holding that AI promises a shining future for all are selling just as much of a bill of goods as the doomsayers. AI--or, better, its antecedent, machine learning--has done some useful things along the way, the authors allow, but on a relatively modest scale: spell-checking, for instance, and advances in medical image processing. Those who buy into the end-of-the-rainbow stuff are courting trouble, they add, such as a lawyer who let ChatGPT write a brief for him that turned out to be so full of holes as to land him in front of a judge. The con of which they write is more comprehensive still, though, based on errant machine-driven financial speculation, data and IP theft, the deprecation of human skills, and other clear and present dangers. A refreshingly contrarian take on AI and the clouds of hyperbole surrounding it. Copyright (c) Kirkus Reviews, used with permission.