In 2013, researchers at the University of Oxford published a surprising number about the future of work: 47 percent of all jobs in the United States, they estimated, were “at risk of being automated” for an undetermined number of years, perhaps a decade or so. two.”
But a decade later, unemployment in the country is at a record low. The tsunami of grim headlines back then – such as “The rich and their robots are about to wipe out half the world’s jobs” – seem completely off target.
But the study authors say they didn’t really mean to suggest that doomsday was near. Instead, they tried to describe what technology was capable of.
It was the first attempt at what has become a long-running thought experiment, with think tanks, business research groups and economists publishing paper after paper to determine how much work is “affected by” or “exposed to” technology.
In other words, if the cost of the tools were not a factor, and the only goal was to automate as much human labor as possible, how much work could technology take over?
When the Oxford researchers, Carl Benedikt Frey and Michael A. Osborne, conducted their study, IBM Watson, a question-answer system powered by artificial intelligence, had just shocked the world with “Jeopardy!” For the first time, test versions of self-driving vehicles drove on the road. Now a new wave of research follows the emergence of tools that use generative AI
In March, Goldman Sachs estimated that the technology behind popular AI tools like DALL-E and ChatGPT could automate the equivalent of 300 million full-time jobs. Researchers from Open AI, the maker of those tools, and the University of Pennsylvania found that 80 percent of the U.S. workforce could see an effect on at least 10 percent of their jobs.
“There’s tremendous uncertainty,” said David Autor, a professor of economics at the Massachusetts Institute of Technology who has studied technology change and the labor market for more than 20 years. “And people want to provide those answers.”
But what exactly does it mean to say that, say, the equivalent of 300 million full-time jobs could be impacted by AI?
It depends, said Mr. Autor. “Afflicted can mean made better, made worse, gone, doubled.”
A complicating factor is that technology tends to automate tasks, not entire professions. For example, in 2016 Geoffrey Hinton, a pioneer in the field of artificial intelligence, considered new ‘deep learning’ technology that can read medical images. He concluded that “when you work as a radiologist, you’re like the coyote who’s already over the edge of the cliff but hasn’t looked down yet.”
He gave it five years, maybe ten, before algorithms would “do better” than humans. What he probably overlooked is that reading the images is just one of many tasks (30 according to the US government) performed by radiologists. They also do things like “consult with medical professionals” and “provide counseling.” Today, some in the field are concerned about an impending shortage of radiologists. And Mr. Hinton has since become an outspoken public critic of the same technology he helped create.
Mr. Frey and Mr. Osborne calculated their 47 percent percentage in part by asking technology experts how likely it was that entire professions such as “telemarketer” or “accountant” would be automated. But three years after publishing their article, a group of researchers from the ZEW Center for European Economic Research, based in Mannheim, Germany, published a similar study that evaluated tasks — such as “explaining products or services” — and found that only 9 percent of the professions in 21 countries can be automated.
“People like numbers,” said Melanie Arntz, the lead author of the ZEW paper. “People always think the number has to be solid somehow, you know, because it’s a number. But numbers can be really misleading.”
In some scenarios, AI has essentially created a resource, not a complete job replacement. You are now an excavator who can use a backhoe instead of a shovel. Or a nurse specialist with access to better information for diagnosing a patient. You may have to charge more per hour because you get a lot more done.
In other scenarios, the technology replaces your work rather than complementing it. Or change your job from one that requires special skills to one that doesn’t. That probably won’t go over well with you.
In both cases, says Mr. Autor, technological developments throughout history have tended to primarily affect wages and wealth distribution – not how many jobs are available. “These types of exercises run the risk of missing the forest by focusing on one very prominent tree,” he said of studies looking at how much human work could be replaced by AI
What he sees as another major focus – how artificial intelligence will change the value of skills – is difficult to predict, because the answer depends in part on how new tools are designed, regulated and used.
Take customer service. Many companies have transferred the task of answering phones to an automated decision tree, where the human operator is only involved to solve problems. But one Fortune 500 enterprise software company has approached the problem differently. It created a generative AI tool to give the agents suggestions on what to say – informing people and their ability to read social cues. When researchers at Stanford and MIT compared the performance of groups who received the tool to those who did not, they found that the tool significantly improved the performance of lower-skilled officers.
Even if a job becomes fully automated, the fate of displaced workers will depend on how companies decide to use technology in new types of work, especially work we can’t yet imagine, said Daron Acemoglu, a professor at MIT and an author of “Power and Progress: Our Millennial Struggle for Technology and Prosperity.” These choices include whether to fully automate work or use technology to augment human expertise.
He said the seemingly frightening numbers predicting how many jobs AI could cut, though it’s not clear how, were a “wake-up call”.
He believes people can “steer in a better direction,” he said, but he’s not optimistic. He doesn’t think we’re on a “pro-human” path.
Any estimates for how much work AI could take over depend very much on humans: the researchers making the assumptions about what technology can do. Mr. Frey and Mr. Osborne invited experts to a workshop to assess the likelihood of professions becoming automated. More recent studies rely on information such as a database that tracks AI capabilities created by the Electronic Frontier Foundation, a nonprofit digital rights organization. Or they rely on employees using platforms like CrowdFlower, where people perform small tasks for money. The employees score tasks on factors that make them susceptible to automation. For example, if it’s something with a high tolerance for errors, it’s a better candidate for a technology like ChatGPT to automate.
It’s not about the exact numbers, say many researchers involved in this kind of analysis.
“I would describe our methodology as almost certainly exactly wrong, but hopefully correct,” says Michael Chui, an AI expert at McKinsey who co-authored a 2017 white paper suggesting that about half the work and 5 percent of professions could be automated.
What the data describes is in some ways more mundane than is often assumed: big changes are coming and it’s worth paying attention to.
