Picture: Pat David / Flickr

As an economist, my own research agenda is based on the premise that guiding AI to beneficial results is the most important problem facing our generation. Since Daron Acemoglu and I are cognate thinkers on this subject, I will take this opportunity to emphasize and extend its core message.

Acemoglu is right on three key points. First, core AI technologies are indeed advancing rapidly and becoming more and more powerful. Improvements in machine learning, in particular, such as the deep learning techniques that have made such rapid progress on the Imagenet dataset– affect more and more of the economy. In good applications, the gain can be big. Adoption is still in its early stages – only 1.3% of businesses in the US adopted robotics, for example, but the numbers are increasing rapidly. Second, the societal implications of AI run deep, especially with regard to the future of work and our individual freedoms and democracy. Third, and most important, the results are not predetermined.

This last point should not be taken for granted. The most common question people ask when I talk about AI is a version of: what will AI do for society? But that’s not the right question to ask; this erases our agency. Acemoglu deserves special credit not only for diagnosing the challenges created by these technologies, but also for suggesting a set of specific solutions. (In fact, Acemoglu has long been one of the strongest and most rigorous advocates of the idea that we can and should lead the course of technical change. item making this argument almost twenty years ago.)

Regarding the diagnosis of Acemoglu, I would like to highlight the following details. When it comes to the effect of AI on the workforce, the real challenge is wages, not jobs. While employment has grown over the past forty years, the real wages of Americans with a high school diploma or less have fallen. Tyler Cowen and others have argued that this is a testament to a lack of technological progress, but that overall GDP and GDP per capita have also increased, and 2019 has seen a record number of billionaires. Building on the work of Acemoglu as good as David Autor, Lawrence Katz, Melissa Kearney, Frank Levy, and Richard Murnane, Andrew McAfee and I argued that technological advances were not incompatible with falling wages for some or even a large portion of the workforce.

The magnitude of these changes, and most important yet to come, is enormous. The value of all human capital in the United States – the sum of the skills, experience, education and know-how of American workers – is likely about $ 240 trillion. This implies that if our decisions altered the trajectory of the effects of technology on the U.S. economy enough to cause a change of even 10% of that value, it would be worth more than a full year of GDP (currently $ 21 trillion). of dollars).

As for the effect of AI on democracy, we should be concerned about increasing polarization and activating Orwellian levels of surveillance. Like Marshall van Alstyne and I have written, it is precisely because digital technologies allow us to better find content and people we love that they can also separate and polarize us. These technologies can also massively amplify the state’s power to monitor the words and actions of its citizens, giving it the power not only to silence critics, but even to shape their thoughts. The implications are increasingly recognized by world leaders; like even Vladimir Putin to put “Whoever becomes the leader in this sphere will become the leader of the world.” In the wrong hands, the result may be what Jean Tirole calls digital dystopia.

But no result is inevitable. We have the ability to lead AI just like we can direct other types of technical changes. Let me focus on three groups that can and should play a role in shaping AI for good: technologists, managers, and policy makers.

“It’s remarkable,” writes Acemoglu, “how much AI research still focuses on applications that automate work.” It is an underestimated problem. While it may be profitable to automate jobs, thereby substituting technology for human labor, in the long run the greatest gains come from the complementarity of humans and the ability to create value in new ways. Moreover, when technology takes the place of labor, pitting humans against machines, it tends to lower wages and lead to a greater concentration of wealth.

In contrast, when technology complements work, wages tend to rise, creating more widely shared prosperity. (In addition to replacing or supplementing the work, Tom Mitchell and I describe four additional considerations about how technology will affect wages: price elasticity, income elasticity, labor supply elasticity, and business process redesign. In many cases, the net result of these six factors will be higher wages.) This is why McAfee and I have argued that “in medicine, law, finance, retail , manufacturing and even scientific discovery, the key to winning the race is not to compete with machines but to compete with machines. “Indeed, during a major conference on AI three years ago, I directly called on the technologists gathered to reorient their work from the replication and automation of human labor to its increase.

Fortunately, a growing number of researchers are working to use AI to augment humans rather than replace them. Take Cresta, an AI start-up that I recommend. While many competitors strive to develop fully automated chatbots that interact directly with potential customers, Cresta is keeping one person up to date. The system works alongside human operators, looking for opportunities to suggest ways to improve the dialogue – suggesting a product or service upgrade, offering a price reminder or coaching on tone and tactics. Via a series of A / B tests, Cresta find that this approach has created demonstrable benefits for clients and also appears to benefit newer and less skilled workers, in particular, by helping to close the wage gap and reduce inequalities.

Managers, entrepreneurs, workers’ representatives and other business leaders also have a key role to play. Like technologists, they too often look at existing processes and ask the easy question: How can machines do what humans do now? The most difficult but ultimately the most valid question is different: how can technology and people work together to create new sources of value? The more powerful and general the technology, the more important it is to rethink the work. As Paul David, Warren Devine, Jr., and others have documented, significant electricity productivity gains in manufacturing did not emerge until managers fundamentally reinvented the way factories were organized, a process that took thirty years or more – long enough to that a generation of managers retire and be replaced by thought. Likewise, modern businesses are subject to a similar dynamic, creating a lull in productivity while intangible investments in organizational and human capital are created in addition to new technologies such as AI.

Policymakers can help each of the first two groups make better decisions by changing the incentives. Take taxation. A key lesson from public finances is that we tend to get less of what we tax the most. The current US tax system treats capital more favorably than labor. If two entrepreneurs each have a billion dollar idea for using AI, the one who employs the most labor is likely to be taxed more than the one who is more capital intensive. To the extent that labor income is more widely distributed, this element of our tax system discourages shared prosperity. This is a powerful argument for leveling the playing field. In fact, there is also a good argument that we should go further as far as we believe there are positive externalities to employment. (2015 book by Robert PutnamOur Children: The American Dream in Crisis describes the negative effects of unemployment, while the 2020 study by Anne Case and Angus DeatonDeath of despair and the future of capitalism documents the increase in deaths from suicide, drug addiction and alcoholism in demographic groups most affected by declining demand for labor.) Depending on the strength of these externalities, they could reverse results suggesting that taxes on capital should be lower than taxes on labor.

This list of actors of change is of course not exhaustive. Economists also have a role to play in guiding the debate, as do the general public. As the power of AI increases, our values ​​become more and more important. It is up to each of us to think deeply about the kind of society we want. It is essential to bring these issues to the forefront of popular discussions.

In the face of all these possibilities for change, I remain a conscious optimist. Acemoglu notes that we are far from a consensus on how to move forward. If it is a challenge, it is also an opportunity to forge a common vision. But our window is short. If wealth and power become more and more concentrated, and if democracy is further weakened, we will reach a point of no return. We can and must act now to prevent this from happening and to redirect AI for the good of many, not just a few.