Silicon Valley and the advances in technology associated with it have become symbolic of American innovation and entrepreneurship. Larger-than-life figures like Steve Jobs and Mark Zuckerberg brought products to consumers around the world—products that they love and which make it possible for us to connect better with each other. Intel has produced chips that make our products “think” faster—do calculations faster—than the best brains in the world. Artificial intelligence (AI) can now beat humans not only in simple games like chess, but in more complicated ones like Go, where the number of possible moves is greater than the atoms in the universe.1 Bill Gates, it would seem, illustrates the best of the American spirit—having accumulated an estimated $135 billion, he began givi“ing massive amounts to charity, as he used his energies to fight diseases around the world and attempted to improve education in the United States.
And yet, for all these virtues, there is a darker side to all of these advances. They create legitimate concerns about job loss. Further, the new industries are prone to numerous abuses, from market power, to invasions of privacy, to political manipulation.
Full Employment in a Hi-Tech World
There is great angst about the job market. In the twentieth century, we created machines that were stronger than humans. Now, we’re able to make machines that are more efficient than humans in routine jobs. AI now presents an even greater challenge to humans. We can make machines that not only perform programmed tasks better than humans but also learn better, at least in certain domains.
Thus machines can outperform people in many key jobs. Better education and job training for workers may be a short-term palliative for many, but computers can and are replacing radiologists, so not even a doctor’s degree provides a safe harbor. It is anticipated that within a few years, self-driving cars and trucks will replace drivers; if true, this is of especial concern, because truck driving today represents a very large source of employment for men who have a high school diploma or less.
The worry is that these labor-replacing machines will drive down wages, especially of low-skilled workers, and increase unem“ployment. The natural answer has been to increase workers’ skills. But in many areas, this won’t suffice: with AI, robots can learn complicated tasks more quickly and perform better than even well-educated humans.
There are those who say not to worry: Look to the past. Markets always created jobs, as the economy restructured. Besides, these techno-optimists claim, the pace of change has been exaggerated. Indeed, it doesn’t even show in the macro-data: productivity increases in recent years are significantly lower than in the 1990s, and in the decades after World War II. Robert Gordon of Northwestern University in his bestselling book The Rise and Fall of American Growth: The US Standard of Living Since the Civil War argues that the pace of innovation has actually slowed.2 Yes, we have Facebook and Google, but these innovations pale in comparison with the importance of electricity, or even indoor toilets and clean water that played such an important role in improving health and longevity.
These past experiences may, however, not be a good guide for the future. More than a half century ago, John von Neumann, one of the leading mathematicians of the mid-twentieth century, suggested that there might be a point where it becomes less expensive to produce a machine to replace a human than to hire and train a human.
These machines will, in turn, be produced by other machines that learn how to produce them. What matters to firms’ decisions to use machines instead of human workers is not just the increase in productivity but also the relative ease and cheapness with which the right machine can be designed, manufactured, and managed. Machines, for instance, don’t go on strike.
One doesn’t need a human resources department to make sure that they are not disgruntled. Machines are unencumbered by emotions. Von Neumann’s forecast has already been realized for certain tasks; as we noted, machines can already outperform radiologists. But the range of tasks and the amount of job replacement may accelerate rapidly, given the advances in AI in just the last five years.
Excerpt From: Joseph E. Stiglitz. “People, Power, and Profits.”