By Gary Saul Morson
"A properly regulated system of AI-powered choice engines could produce massive welfare benefits," concludes Cass Sunstein in "Imperfect Oracle, " his study of what artificial intelligence can do for humanity. "It could make life less nasty, less brutish, and less short -- and less hard." Many people today see great potential in large language models and other, more ambitious, AI applications. But what does he mean by "AI-powered choice engines"?
Mr. Sunstein, a professor at Harvard Law School and a former member of the Obama administration, identifies the real benefit of AI as its capacity to overcome human "cognitive biases." Deeply influenced by the field of behavioral economics, he argues that people tend to value avoiding losses rather than pursuing equivalent gains, pay too much attention to the examples of outcomes that are most familiar to them, and then to be "unrealistically optimistic." They use "heuristics" that humans evolved for making snap decisions but that can mislead them at other times. "People tend to focus on the short term, not the long term, " he notes. We trust our intuitions when we should rely on rational calculation. "Intuitions and impressions should be replaced by computations," Mr. Sunstein concludes.
Unlike some other devotees of AI, Mr. Sunstein recognizes its limitations. The economist Friedrich Hayek maintained that a centrally planned economy could not work because the number of relevant and changing facts that the planners would need to know and understand is simply too large. The needed information, Hayek observed in a passage Mr. Sunstein quotes, "never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate people possess." As a result, prediction faces obstacles that more computing power cannot overcome. "It is high time . . . that we take our ignorance more seriously," Hayek memorably observed. Mr. Sunstein adds: "We need to take AI's ignorance more seriously" too. When it comes to forecasting what a society, or even an individual, will do decades from now, AI is, and will remain, no more knowledgeable than the central planners of the past century.
What's more, Mr. Sunstein acknowledges, there are some decisions that people should not want to leave to AI choice engines. Sometimes we must make our own errors, if our lives are to be ours. Life is not only the experiences we undergo, but the actions we undertake, actions that are based on choices we are free to decide one way or another: to take a job or leave one, to get married or have a child, to get divorced.
For all his caution about AI, however, Mr. Sunstein expresses firm confidence that wise "choice architects" would create "properly regulated" choice engines and allow their "ideal form" to make wiser decisions than people usually do. Time and again, no sooner does Mr. Sunstein point to a limitation of AI than he finds a way around it. Sometimes his loopholes swallow up the whole. To be sure, autonomy matters, he allows, but sometimes "private and public institutions" should override our preferences, so long as they do so "because of some kind of error on the part of the choosers, perhaps in the form of a lack of information, perhaps in the form of some kind of behavioral bias." And when is that not arguably the case? As Mr. Sunstein describes human decision-making, a devoted official will always be able to find some such error.
A keen insight about the importance of choice, followed by an argument to justify yielding that choice to the "choice designers": that is the pattern of this book. We often forget, Mr. Sunstein shrewdly points out, that we not only choose, we also choose whether to choose. These second-order choices explain why people sometimes prefer to leave their decisions to others. Don't we frequently trust experts to know more than we do? Isn't that what we long to do?
Dostoyevsky argued that people frequently desire to surrender their freedom to avoid guilt and regret; they will readily do so provided they are assured that they are really gaining a "higher" freedom. That is what socialists usually say. Mr. Sunstein argues that the real deprivation of freedom is to insist that people choose when they would rather choose not to choose. He calls this insistence "choice-requiring paternalism," a phrase meant to throw the charge of paternalism back at the libertarians.
But earlier in the book, Mr. Sunstein has pointed out the obstacle to his own plan. Humans do not want to be given happiness. We want to achieve it; or, as Dostoyevsky once observed, "happiness lies not in happiness but only in its pursuit." This idea shaped 20th-century dystopian novels, and Mr. Sunstein quotes a crucial passage from Aldous Huxley's "Brave New World," in which a character objects to the decisions made on his behalf to optimize his happiness: "But I don't want comfort. I want God, I want poetry, I want real danger, I want freedom, I want goodness. I want sin." To surrender such decisions to some putatively wiser entity is, in some essential sense, to surrender our humanness.
Beyond this, Mr. Sunstein underestimates the eagerness of officials to expand their power ("administrative ecstasy," as Dostoyevsky called it) and their readiness to believe they are wiser than the error-prone people they manage. For all his acknowledgments about the importance of human autonomy, I finished this book with the very phrase "choice architect" making my blood run cold.
--Mr. Morson, a professor of Slavic languages and literatures at Northwestern University, is the author of "Wonder Confronts Certainty: Russian Writers on the Timeless Questions and Why Their Answers Matter."
(END) Dow Jones Newswires
December 15, 2025 14:01 ET (19:01 GMT)
Copyright (c) 2025 Dow Jones & Company, Inc.
Comments