Automate This! How AI technology could disrupt nearly half of all jobs — and international politics
Welcome to the first instalment of 'Automate This,' a Day 6 series about the future of work in an artificially intelligent world.
Want to know whether your job could be on the brink of automation? Try asking an artificially intelligent algorithm.
That's what Michael Osborne did back in 2013 — and the results were staggering.
Along with his colleagues, Osborne built an algorithm that predicted as many as 47 per cent of U.S. jobs could be susceptible to automation.
The algorithm didn't hold back from predicting the demise of high-paying, white-collar jobs — including, among others, accountants and economists.
But when it listed fashion models among the ranks of the automatable, even Osborne was skeptical. At first.
"You might naively think that this is a non-automatable occupation, but if you look at the characteristics of the job, in fact it does seem relatively routine," he says.
"It's not entirely inconceivable that in the not-too-distant future, we might be able to use computer-generated [fashion] models, for at least some purposes — for the purposes of advertising, for instance."
Osborne is a machine learning researcher and co-director of the Oxford Martin Programme on Technology and Employment.
Four years after his original study made headlines, he says some of his predictions have started to come true — with significant political repercussions.
"Living within the world of machine learning, I think you'd have to say the advances in even the last five years have been startling."
Canadian workers on the brink
In recent decades, machine learning and A.I. technologies have mastered an incredible range of non-routine tasks — from navigating busy streets to giving legal advice.
According to the Brookfield Institute, existing technology could already automate 42 per cent of the tasks that Canadian workers do at their jobs, leaving nearly half of Canadians at a high risk of disruption in the next decade or two.
Living within the world of machine learning, I think you'd have to say the advances in even the last five years have been startling.- Michael Osborne
Steve Kelly is one such worker.
Kelly is a heavy equipment operator with SunCor in Fort McMurray. He's concerned that A.I. technology could put him and his fellow employees out of a job.
"Depending how an automated fleet rolls out with driverless trucks, it's hundreds to thousands of people. We're talking two, three, four thousand people, just as haul truck drivers, that could potentially lose their positions," says Kelly.
"If I were to lose this position today … I don't know what would happen. I'd have to look at changing careers, paths — it changes everything."
Researchers at the University of Toronto's Mowat Centre say anywhere from 1.5 million to 7.5 million Canadian workers could lose their jobs to automation in the next decade.
Osborne believes the real number could be even higher.
"It's possible due to the enormously rapid advances in technology such as machine learning and robotics that are already having an impact on the world of work," says Osborne.
It's not all bad
The Canadian government is investing millions of dollars in artificial intelligence research through organizations like the Vector Institute, an independent not-for-profit research centre housed in Toronto's MaRS Discovery District.
Proponents of A.I. technology say it will attract investment and create new jobs while enhancing productivity and boosting economic growth.
"Even if only some of the tasks within a job are automatable, the skills required to complete that job may change completely."- Michael Osborne
Krista Jones, managing director of Work and Learning at MaRS, says there are plenty of reasons to be optimistic about the workplaces of the future.
"When I look at what A.I. is going to do for the medical community, when I see what it can do to help us with data and some of the analysis on the environmental side, I get really excited about the use of big data and algorithms to help us solve some big problems that we as society face," Jones says.
"In a more practical sense, [I] really am excited about what it's doing to help workers prepare for the future."
Jones believes A.I. technology could one day be deployed to help build personalized retraining programs for workers, helping to ease the technological transition.
Other researchers argue that while many of the individual tasks involved in certain jobs may be automated, most jobs will evolve rather than disappear completely.
But Osborne says even those changes will have huge implications for workers.
"This improvement in productivity resulting from technological change is not something we should dismiss. We need that productivity growth," he says. "Ultimately, that will yield the wealth that will enable us to tackle some of the social challenges we're facing.
"But nonetheless … we shouldn't lose sight of the fact that even if only some of the tasks within a job are automatable, the skills required to complete that job may change completely.
"Those are not necessarily occupations we should identify as immune to the negative effects of automation."
Politics on the brink
If workplace automation continues at the current rate, we could be headed for significant political upheaval, Osborne says.
In fact, our democracies are already starting to feeling the impacts of technological change.
Earlier this year, Osborne's colleagues published a study that found automation played a role in the election of U.S. President Donald Trump.
"They found a direct link between the fraction of work in a county that was potentially susceptible to automation, and the fraction of people in that county that voted for Trump."
The correlation held even when the researchers controlled for factors including voters' exposure to globalization and immigration.
"Interestingly, they even corrected for Republican voting," says Osborne. "Counties that had more routine work were more likely to vote for Trump than they had been for Mitt Romney."
It's not the first time technological advances have lead to radical political change.
In the early decades of the Industrial Revolution, workers did not share in the enormous economic gains associated with new technologies like the steam engines, Osborne says.
Counties that had more routine work were more likely to vote for Trump than they had been for Mitt Romney.- Michael Osborne
"Some findings say that it took as long as 70 years for the average worker to see any improvements in pay at all — and in fact, in the initial 40 years, wages actually declined for most workers."
Those discrepancies famously led workers to revolt against the new technologies, as the Luddites did in the early 19th century.
"The concerning parallels here are that perhaps it's the impact of automation on low-paid workers in the U.S. that has led to some of those workers voting for the outsider candidate, the now-President Donald Trump."
It's possible that this situation could repeat itself around the globe, Osborne says, noting other researchers have found that many countries whose workforces are most at risk from automation are also the most politically unstable.
"This is far from a causal story," says Osborne. "[But] as technologies are pushed even into those developing nations that are relatively unstable, more instability may be the predominant result."
Click here for Part Two, Three and Four of Automate This!, our special series about the future of work in an artificially intelligent world.
For the full interview with Michael Osborne, download our podcast or click the 'Listen' button at the top of this page.