March 30, 2017
Alexander Panetta THE CANADIAN PRESS
WASHINGTON , DC—Should we be preparing for the coming invasion of job-stealing, career-crushing robots? It’s a question that’s moved from science-fiction novels to the tip of policy-makers’ tongues.
Canadian and American policy-makers have just delivered very different answers.
Canada’s finance minister tabled a budget that mentioned artificial intelligence and skills training dozens of times, with entire sections on each subject and $5.2 billion for worker re-training.
His U.S. counterpart: not so worried about a wave of job-killing automation.
“It’s not even on our radar screen,” U.S. Treasury Secretary Steve Mnuchin told the Axios website last week. “(It’s) 50-100 more years (away)… I’m not worried at all,” he said, adding: “In fact, I’m optimistic.”
The Trump administration’s budget proposal didn’t specifically mention skills, automation, or artificial intelligence a single time.
So who’s right?
Silicon Valley reacted with swiftness and stupefaction. Some of the world’s most famous tech billionaires are expressing alarm about rapid change approaching, starting with automation potentially wiping out millions of driving jobs.
The headlines in the tech press were scathing. Wired magazine wrote: “Hate to Break It to Steve Mnuchin, But AI’s Already Taking Jobs.” TechCrunch said: “Steve Mnuchin has been compromised (by robots).”
Jerry Kaplan, who lectures at Stanford University on the social, economic, and legal questions raised by AI, said: “I think Mnuchin’s comments regarding AI are ill-informed, an unfortunate hallmark of the new U.S. administration. He may not be concerned about it, but just about anyone who has a real job and does productive work should be.”
Yet the picture isn’t exactly black and white.
First of all, those few comments from Mnuchin don’t encapsulate the full spectrum of attitudes in Washington. Just this week, the White House announced the creation of the Office of American Innovation.
The new office will be led by President Donald Trump’s son-in-law—in addition to being tasked with the Middle East peace file, Jared Kushner will now lead an office with a half-dozen different loosely related mandates including: developing a “workforce of the future,” veterans’ services, and the opioid epidemic.
Congress clearly sees big changes coming.
Two committees are preparing for self-driving cars in the House of Representatives—the commerce committee is considering manufacturing standards, and the infrastructure committee is contemplating next-generation roads.
The infrastructure chair described Wednesday being mesmerized by a self-driving SUV. Bill Shuster even defended the technology as safe—when an accident occurs during testing, as happened with Uber days ago, he said it’s usually caused by a human error in another vehicle.
“We went 35 miles (in our test). It was incredible,” Shuster told a panel event.
“It will change the way we drive. It will help improve, or reduce, congestion… We’re talking about tremendous savings in lives, in cost, in damage.”
Yet it might also cost millions of driving jobs.
What happens to truck-drivers? Or to restaurant workers, for that matter—even Trump’s first pick for labour secretary, who ran fast-food chains, publicly mused about the benefits of replacing his workers with robots.
It’s not so cut-and-dried there either.
Estimates of the job impact range broadly. While a famous Oxford study said almost half of human jobs were at risk of automation, a more granular-level McKinsey study suggested most jobs will still exist—but half of specific tasks will disappear within 20 to 50 years.
This week the U.S. government published a study that attempted to put even more precise numbers to it.
The conclusion: robots steal human jobs, but not many.
The research for the U.S. National Bureau on Economic Research found that one robot per 1,000 workers reduced the employment population by 0.34 percentage points, and salaries by 0.5 per cent.
It then estimated the future impact, in a world with more robots.
Under two scenarios, one with a conservative estimate for robot-proliferation, and a more dramatic scenario, it estimated employment rates would fall between 0.54 and 1.76 percentage points, while wages fell 0.75 to 2.6 per cent.
Frank Pasquale takes a middle-ground view.
The Yale fellow and University of Maryland professor has written extensively about the legal, ethical, and social challenges of AI. He points to encouraging research, about new jobs replacing old ones.
For example, James Bessen has written about how ATM machines haven’t replaced bank tellers. On the contrary. The Boston University researcher charts an increase in bank employment—workers have stopped handing out cash, and now sell financial products. He finds similar results for paralegals and supermarket checkout clerks.
David Autor of MIT uses the example of car windshields: they’re now installed by robots, he says, but humans remove the broken ones, fix them, and fit in replacements.
Mnuchin should be a little more worried, he says. Not panicked, but concerned about the transition, and the many humans whose livelihoods could be sideswiped by fast-approaching change. That especially applies to people who make less than $20 an hour, he said.
“I think that is too sanguine a view (from Mnuchin),” he said.
“The bottom line: we have to have a robust and substantive policy for transitioning people in areas that are imminently going to be automated.”