When the University of Chicago asked a panel of leading economists about automation, 76 percent agreed that it had not historically decreased employment. But when asked about the more recent past, they were less sanguine. About 33 percent said technology was a central reason that median wages had been stagnant over the past decade, 20 percent said it was not and 29 percent were unsure. Perhaps the most worrisome development is how poorly the job market is already functioning for many workers. More than 16 percent of men between the ages of 25 and 54 are not working, up from 5 percent in the late 1960s; 30 percent of women in this age group are not working, up from 25 percent in the late 1990s. For those who are working, wage growth has been weak, while corporate profits have surged. "We're going to enter a world in which there's more wealth and less need to work," says Erik Brynjolfsson. "That should be good news. But if we just put it on autopilot, there's no guarantee this will work out."
Now, two physicists have shown that one form of deep learning works exactly like one of the most important and ubiquitous mathematical techniques in physics, a procedure for calculating the large-scale behavior of physical systems such as elementary particles, fluids and the cosmos. The new work, completed by Pankaj Mehta of Boston University and David Schwab of Northwestern University, demonstrates that a statistical technique called "renormalization," which allows physicists to accurately describe systems without knowing the exact state of all their component parts, also enables the artificial neural networks to categorize data as, say, "a cat" regardless of its color, size or posture in a given video.
"They actually wrote down on paper, with exact proofs, something that people only dreamed existed," said Ilya Nemenman, a biophysicist at Emory University.
But Watson is still a work in progress. Some companies and researchers testing Watson systems have reported difficulties in adapting the technology to work with their data sets. "It's not taking off as quickly as they would like," says Robert Austin. "This is one of those areas where turning demos into real business value depends on the devils in the details. I think there's a bold new world coming, but not as fast as some people think." IBM needs software developers to embrace its vision and build services and apps that use its cognitive computing technology. In May of this year it announced that seven universities would offer computer science classes in cognitive computing and last month IBM revealed a list of partners that have developed applications by tapping into application programming interfaces that access versions of Watson running in the cloud. Big Blue said it will invest $1 billion into the Watson division including $100 million to fund startups developing cognitive apps. "I very much admire the end goal," says Boris Katz, adding that business pressures could encourage IBM's researchers to move more quickly than they would like. "If the management is patient, they will really go far."