The real danger lies not in AI, but in ourselves.
Brave New Technology
In the early 1990s, I had the very good fortune to work on Capitol Hill for then-US Senator Sam Nunn (D-Georgia). As an elected official and a boss, Senator Nunn was widely known for seeking balanced perspectives on policy problems. This wasn’t a tic or the narcissism of small policy differences. He recognized that many issues were 51–49 propositions, meaning that it was often hard to tell which side was which. “Get enough of those wrong,” he said, “and they really start to add up.”
I was reminded of that wisdom while reading Marco Magnani’s Making the Global Economy Work for Everyone: Lessons of Sustainability from the Tech Revolution and the Pandemic. Magnani, an economist at the University of Rome, has produced a readable and concise introduction to trade-offs in the new age of digital automation. If he’s right, the next few decades may turn out to make all previous economic transitions (farm to factory, factory to services, services to information, and information to globalization) look tame by comparison.
One of Magnani’s important contributions is to provide a glossary of terms for what he calls a “tsunami” of new technologies that are in the beginning stages of rewiring our economic system. For anyone, like myself, who’s ever struggled to get their mind around concepts like “blockchain,” “quantum computing,” “artificial intelligence,” or the distinction between “machine learning” and “deep learning,” reading the first few chapters is a painless way to stem one’s own ignorance on technology-related topics, and worth the investment of time and treasure. But it is only the beginning of the value his insights provide.
The more important aspect of Magnani’s analysis is in the way he thinks about the radical uncertainty this flowering of new technology is creating. It’s helpful in this regard not to think about any of the hot new technological “toys,” like, say ChatGPT, which threatens to upend Google’s dominance of the search engine sector, but as a dense constellation of related digital technologies that will interact and recombine in unanticipated and unplanned ways. Much of this will be good, helpful, and reflective of the natural process of innovation. That’s one half of the 51–49 dilemma. The other half is the challenge these known and unknown technological developments will have on existing economic and social relationships, a problem Magnani refers to collectively as “disintermediation” or the tendency of digital technologies to generate and boost profits by targeting inefficiencies and frictions in the finance, design, production, and distribution of goods and services.
Now, we all love efficiency and the lower costs it brings across the economy. We get more of the products and services we want faster and cheaper. But efficiency contains its own costs. Human participation in many economic activities slows down processes rather than speeding them. People produce value (in fact, I’d argue people ultimately produce all value) but they also get tired and distracted, make mistakes, and generally contribute to economic friction.
Take a hospital visit for instance. The expert nurse who puts in your IV line with minimal pain is immensely valuable. (I know this from recent experience.) A shift change at the hospital, however, is a massive information management problem as doctors and nurses try to communicate the trajectory of care for their patients to a new team. Details get missed, delays ensue, and suffering is exacerbated. So, is the human element valuable in delivering high-touch service or a friction to be automated in the name of efficiency?
The answer, unfortunately, is probably both, as we trade off some efficiency for a unique human benefit. Leaning too far into technology can make our situation worse rather than better depending on what we’re measuring. Moreover, one person’s efficiency improvement often turns out to be another person’s unemployment. Any good economist will sterilize these problems with terms like “in the long-term” and “transition costs.” And, admittedly, the weight of historical evidence is overwhelmingly on the side of the long-term perspective. This can be verified by a simple mental experiment: looking at the totality of your circumstances (economic and social) would you prefer to have lived in an earlier era? The idea of living in a world without anesthesia or antibiotics has always put me decisively on the side of “Now’s good, thanks!”
Even net positive change entails costs. Arriving at our healthier, wealthier, and more comfortable era has meant a lot of work and suffering for the generations who had to live through social, psychological, and emotional upheaval as they moved from small, intimate, face-to-face communities to more isolated, fragmented, and anonymous urbanized environments. Further, these individual, community, and cultural stresses tend to “filter up” into social and political movements that can—and in the last century did—result in considerable violence and death. If we say “yes” to the present, we should try to keep in mind that the costs of its blessings were born primarily by others. Why would we expect turmoil not to be part of the package in our lives?
The Western industrial revolution took two centuries to reach full flower before it began to transition toward services and information over a 20- to 30-year period. The shift from services and information to an integrated global economy (as marked by China’s accession to the World Trade Organization) was 20 years. And we’ve witnessed, in real time, some of the severe dislocations this most recent chapter of economic evolution has caused: concentrated dis-employment in communities that depended on manufacturing jobs, a sudden decline in life expectancy due to addiction, and “deaths of despair,” and social and political turmoil across the developed world that have only made themselves broadly felt in the past 10 years or so.
In other words, as a society and economy, we’ve been through a lot of change in a highly compressed period of time. Magnani tells us the new digital technologies are going to “disintermediate” the entire economy even more quickly, exposing classes of previously “safe” workers to technological unemployment. The rewiring of finance, design, production, and sales through the use of artificial intelligence, “big data,” 3-D printing, and reductions in internet latency will decentralize and redistribute business activities while eliminating layers of middle management that had, heretofore, been devoted to coordinating business processes. If history is a guide, the efficiencies created will, over time, boost productivity and incomes and thus generate new demand to keep boosting demand for labor. Indeed, the principal problem of our economy at the moment is not surplus labor but a chronic shortage of both people and skills. However this process turns out, it won’t be smooth, painless, or easy. There’s a lot of short-run before we get to the long-run.
Magnani goes on to say that our assumptions about the benefits of technology may not turn out to be true this time. The primary beneficiaries of these capital-intensive digital improvements are most likely to be, well, the owners of capital, delivering profits and income to the wealthy while squeezing the work opportunities of everyone else and putting humanity on a treadmill of trying to keep up with microprocessors. (For a fictional discussion of this challenge, I highly recommend Klara and the Sun, Kazuo Ishiguro’s meditation on humanity’s fraught relationship with technology that’s “smarter” than its makers.) Regardless of who’s on the 51 and 49 of this cost-benefit question, the time to start preparing for even relatively short-term or benign disruptions is now.
In Magnani’s telling, the first problem to be addressed is the future of skills, which itself is a multi-dimensional challenge. The most important work skill for the future, he says, is the ability to use digital technology. High-end technologists and designers, supported by big data and various forms of artificial intelligence, will continue to provide technological breakthroughs and adaptations to existing capacities, but the average person’s burden will be in trying to keep up with the way these innovations are reshaping their work.
His principal recommendation for achieving this end is “pre-distribution”: establishing mechanisms that build individually-owned resource endowments for education and training on a life-cycle basis. This idea is similar to “baby bonds,” which representatives of both political parties have shown interest in as a way of helping to level the socioeconomic playing field without some of the work disincentives of other approaches, like a negative income tax or a universal basic income. Magnani is less specific as it relates to reskilling existing workers, but it is easy to imagine structures like a lifelong learning account, jointly funded by contributions from workers, employers, and government. This would build up pools of savings for retraining, relocation, and other reemployment needs that, if not used during a career, could be converted to retirement savings. We cannot guarantee equal outcomes, but policy innovations like these can help improve access to opportunity.
The other skill dimension that Magnani emphasizes is noncognitive or “soft skills.” Data from the past few decades shows the demand for these skills—communication, collaboration, perseverance, teamwork, and others—is rising as the call for skills in repetitive, manual tasks that dominated manufacturing has fallen. Employer surveys regularly cite the shortage of noncognitive skills as the most pressing human resource problem their firms face. Further, while digital technologies can replace a good deal of human cognition and, through robotics, human movements, social-emotional sensitivity, and responsiveness are (currently) beyond the reach of artificial intelligence. (Note that anytime an expert says that AI can’t do thus-and-so, it is always wise to make a mental reservation: yet.) Magnani suggests the digital transition stresses we are undergoing will likely strain these social-emotional capacities further.
Even now, noncognitive skill deficits are one of those things that we all see, but few have concrete policies and strategies for developing them. These skills are not readily susceptible to being turned into formal education and training; they are mostly “caught” rather than “taught” through families, civil society organizations (e.g., churches, youth leagues), and schools. The ongoing fragmentation of American families and communities is probably the headwaters of the noncognitive challenge and represents a serious, difficult-to-overcome constraint on increasing the supply of noncognitive skills. For low-income populations, it may be possible to compensate through certain high-quality interventions like nurse home visiting and rigorous sector-based training that combines technical and interpersonal skill development.
The noncognitive skill challenge is intimately connected to the rise and proliferation of digital technology in another way. As noted above, these types of interpersonal skills are increasingly needed in the nation’s workforce. This trend, Magnani notes, will likely accelerate as routine types of work are automated away, and demand for humane skills rises due to demographic change. Social care—doctors, nurses, home health, elder care, mental health, teachers, pastors, trainers, and the like—will be needed. Moreover, there is likely to be a period of great social stress as the economy reaches a new equilibrium. Simply having more people trained in the skills needed to deliver care might help backstop American society against the fear, despair, and turmoil that comes with socioeconomic transition. As Richard Reeves, a scholar at the Brookings Institution has noted, an important part of this is helping to convince young people, especially boys and men, that these are honorable and viable careers for them and not just their mothers, sisters and partners.
An advanced economy is akin to a living, breathing jigsaw puzzle. Such a puzzle grows and changes in unexpected ways, reconfiguring itself as it goes. The pieces change shape, size, and color and require the people who live within them to grow and adapt. The trick is ensuring that as the puzzle morphs, we solve more problems than we are creating. In the world Marco Magnani describes, the pace of digital change leaves too little time for more analog social primates like human beings to keep up. Our challenge, then, is less to manage (or try to stop) technological development than it is to manage ourselves as we encounter the change. We will need to remember that while this may be hard, the growth and rising living standards it promises will probably bring great benefits to our kids and many generations beyond them. This is America’s intergenerational compact, and it remains a future worth working for.