Superhumans, All-Too Humans, and the Deciding Factor

In March, when an administration’s national-security group chat was accidentally leaked to a journalist, the majority of reactions seemed to fall into one of two camps. On the one side, there were accusations of a failure to protect classified information, in this case a debate over airstrikes against the Iran-backed Houthis. And on the other side, insistence that no war plans were discussed and the entire episode was overblown. My own reaction to the so-called “Signalgate” affair, however, was decidedly different. Reading through the exchange between America’s top security team, I couldn’t help but think, their conversation sounds so…normal. Put simply, the approach to decision-making by our top security officials felt human, all-too human to me.

Of course, the team is made up of humans, and thus their decision-making process—from the factors they weighed to the reasons they yielded and the points they found convincing—will necessarily share much in common with that of their fellow mortals. Which, I suppose, is another way of saying that it was necessarily imperfect. And that imperfection, more than anything else, is a persuasive argument for an embrace of “superagency.”

A term colloquialized by LinkedIn co-founder and recent EconTalk guest Reid Hoffman, “superagency” is what happens when we learn to stop worrying and love AI. Think of it, he says, not as a force of displacement or dehumanization, but instead as force multiplier for human agency. As he tells Russ in a recent episode about what can go right with AI, the revolutionary technology can amplify our intelligence, our expertise, and our decision-making ability. Specifically in the case of the latter, AI can analyze and synthesize vast amounts of information, thus helping us cut through bias and enable better risk assessment. 

Strangely enough, as another EconTalk episode makes clear, Hoffman’s case may actually be bolstered by the rather disastrous introduction of AI into the English Premier League: As ed-tech innovator Daisy Christodoulou explains in her conversation with Russ about the cost of perfection in decision-making, there is such a thing as too much certainty. For proof, one need only look at the rinse-and-repeat process whereby VAR (video assistant referee)’s decisions outrage fans, authorities change the rules to correct the problem, and the now more-complicated rules result in still more outrage. In the end, concludes Christodolu, what we really seek in refereeing—as in all areas of life—is a just-right balance between common sense and consistency, a balance that technology alone is unable to provide. Instead, the answer lies in “comparative judgment,” in which an algorithm combines the sum total of our decisions into a measurement scale. A la Hoffman’s “superagency,” the scale can then allow us to correct for our innate difficulty in judging things in absolute terms.

Indeed, it seems reasonable to conclude that in the (very) near future, every decision, in every area of life, will be improved by our joining forces with technology. Computer scientist and podcaster Dwarkesh Patel even argues in his talk with Russ about the exponential scaling that characterized AI’s evolution to date that if the current rate of progress continues, we’ll arrive at an AI that can take over most human tasks before we know it—and whether we want it or not. This certainly seems to be the conclusion put forward by restaurateur Will Guidara, who spoke with Russ about the secret to truly resonant hospitality. In a recent Instagram post, Guidara wrote that contrary to the widespread belief that no machine could ever replicate the warmth of a human being, AI is actually more hospitable than we are: never tired, never forgetful, completely lacking in ego, and—most important—able to decide how best to make people feel seen, and much faster and better than we ever could.

And then, as I nearly gave up the search for a cranny in which humans alone still reign supreme, Guidara reminded us of his post’s date: April Fool’s Day.

And so, for the time being, then, the jury’s still out on whether AI can improve our decisions, about absolutely everything. For all those interested in the debate—happy to create a group chat.

Marla Braverman

Editor at EconTalk

Mining the Conversation

A selection of additional EconTalk episodes that explore the evidence related to better decision-making, with technology and without.

Cass Sunstein on Infotopia, Information and Decision-Making How can we avoid the trap of information cocoons that reinforce our prior conceptions? Simple: Find strength in numbers. Legal scholar Cass Sunstein’s survey of Internet platforms for aggregating information shows how for every digital echo chamber, there’s a knowledge market for self-correction and the attainment of accuracy.

Phil Rosenzweig on Leadership, Decisions, and Behavioral Economics Insisting that not all decisions are created equal, management expert Phil Rosenzweig analyzes that unique combination of skills—most localized in our left brains—that, together with the “right stuff,” can result in winning calls when the game’s too important to lose.

Annie Duke on the Power of Quitting Instead of a relentless focus on persevering through adversity, former professional poker player Annie Duke wants us to recognize the opportunities inherent in folding—and the opportunity costs associated with sticking to a losing outcome. She explains how society’s conflation of grit with character has made quitting unnecessarily hard, and why our deep-seated desire for certainty impedes our decision-making ability. 

Doyne Farmer on Chaos and Complexity Standard economic models—that is, the ones in which people respond to incentives and maximize utility—might work in theory, but in practice they often break down when faced with complexity. Fortunately, physicist Doyne Farmer has a solution: “complexity economics,” or the use of big-data and supercomputing to help economists build better models, make better predictions, and solve some of the biggest problems facing society. 

Conversation Starters

An eclectic collection of books, films, poems, and podcasts that describe how we make decisions, decision-making at its best (and worst), and when outsourcing decisions to computers both does,
and definitely doesn’t, work.

Thinking, Fast and Slow A classic of behavioral economics, psychologist Daniel Kahneman shows how tapping into both types of thinking—the intuitive and emotional on the one hemisphere, and the logical and deliberative on the other—can save us from bad decisions.   

The Hunt for Red October When Sean Connery—that is, Marko Ramius—appears to go rogue with his country’s ballistic-missile submarine, CIA analyst Jack Ryan must prove to his trigger-happy superiors that the Russian submarine captain intends to defect to the U.S. Along the way, we’re treated to a deep dive into issues of uncertainty and risk in the security realm.

Moneyball In this biographical sports drama based on the book by the same name, the Oakland A’s general manager uses computer-generated analyses to identify undervalued talent and acquire new team players for half of what his rivals might pay. The approach naturally triggers skepticism and resentment—until it scores a homerun in the World Series.

“The Love Song of J. Alfred Prufrock” A lyrical lesson in the dangers of inaction and indecision, T.S. Eliot’s famous poem shows that an obsession with doing it “right”—when courting, choosing a hairstyle, or writing poetry—is almost always a recipe for regret and disappointment. So go ahead, eat that peach.

The Amazing Digital Circus An Australian animated web series based on the 1967 apocalyptic story “I Have No Mouth, and I Need to Scream,” The Amazing Digital Circus—one of the most-watched animation videos in YouTube’s history—follows a group of humans trapped inside a circus-themed virtual reality game, where they are subject to the erratic decisions of their AI ringmasters. Don’t let the cute cartoon characters distract you from the terrifying warning at its core.

Comments? Questions? Email us at [email protected]. We’d love to hear from you.