It comes as a surprise. A message that technology is taking away our jobs and will destroy society can be expected from a Luddite delivered on a carrier pigeon. This message comes from the very bastion of technology, the Massachusetts Institute of Technology (MIT), and was delivered by the New York Times. Never mind that MIT recently issued a report that assured us that “recent fears about AI leading to mass unemployment are unlikely to be realized.”
The change in direction, from automation creating jobs to creating job loss and inequality, has been the growing consensus of economists, including Acemoglu. Our Jim Anderton arrived at the same conclusion. See Artificial Intelligence is Here. Will Anyone Have a Job? in our End of the Line series.
Acemoglu has been trying for some time to walk back a pro-technology message that his own institution, and many others, had been sending: that technology, automation and AI will add jobs overall. Only if you are “low skilled” should you worry. It was a message that was well received, if only because it was heard by those who have embraced and benefitted from technology. Those who have aged out, lost their jobs, were made homeless or have otherwise been marginalized by technology’s relentless advances are probably not reading the New York Times or the Harvard Business Review, or getting press releases from MIT.
But looking at the data, many academics, including Acemoglu, have reconsidered the rosy predictions of technology granting full and gainful employment for the worthy (that’s you and me, of course), even if it left those who couldn’t or wouldn’t keep up in its wake. For the first time, we are hearing of an emerging and disturbing trend—that the jobs that technology has enhanced or created will suffer.
Of his earlier work, where Acemoglu, praised technology as a magic ingredient that makes for the wealth of nations, he now says he has reconsidered.
“It’s too restrictive a way of thinking,” he said. “I should have been more open-minded.”
As he continued to study the data, a new and disturbing trend appeared. Technology was not the magic ingredient but may rather be a poison. The social disparity and job displacement caused by technology (AI, automation and software in particular) became apparent.
In a study published last year, Acemoglu and collaborator Pascual Restrepo, an economist at Boston University, attribute at least half of the increase in wage disparity to the adoption of technology. The researchers conclude that workers’ wages decrease as spending on machinery and software increases. This contrasts with the “golden age” from 1950 to 1980, when technology pickup increased and workers flourished.
“It’s made me less optimistic about the future,” said Acemoglu.
Technology Giveth With One Hand …
Consider Uber, once the darling of Wall Street. Uber’s valuation at its 2019 IPO was expected to be higher than GM, Ford and Fiat combined. Uber’s technology created the gig economy. At its height, before the pandemic, the San Francisco ride-sharing company had 5 million drivers worldwide. But simultaneously, Uber—along with several tech companies—were hatching plans for autonomous vehicle technology and replacing human drivers with self-driving “robo-taxis.”
Uber is certainly not the first to convince us that technology makes jobs. After the Luddites burnt down the weaving machines that cost them their jobs, industrialists learned to sell technology on the jobs it would create. We are now accustomed to embracing technology for the better, higher-paying and more fulfilling jobs it will create. Honestly, do you want your kid to design the robot creating the spot welds or working on the assembly line? Have your kids picked up Python (a programming language favored by AI developers)? Indeed, a degree in computer science with an emphasis in AI is the golden ticket of our times. An AI and machine learning (ML) developer could expect a salary ranging from $150,000 to $175,000 for new grads two years ago, as reported by the Wall Street Journal.
Similarly, we have generative design, a technology that could theoretically assist the engineer. Generative design may still be in its infancy—still learning to crawl—but it gets smarter every day. We are prone to dismiss generative design as an assistant and the idea that it could replace a design engineer seems preposterous.
But even as we assure ourselves that our jobs are beyond the capabilities of machines, each wave of technology takes away more jobs. We may have conceded manual labor to machines, but with AI, more of the jobs are office jobs. IBM’s Watson beat humans, including all-time Jeopardy! champion, Ken Jennings, spawning natural language computer responses that were to replace almost everyone you expected to pick up the phone. Replacing the switchboard operator with a voice menu came early. AI technology has replaced the airline agent, the nurse, the banker.…
We will not go willingly. For the AI that pretends to write news briefs, we insist that our own are better. For those who push generative design for parts, we will insist that only our parts are manufacturable and dependable. At the same time, we watch AI learning and getting better. How much time do we really have?