The job-destroying potential of artificial intelligence and automation has created much angst around the fast-evolving technology.
But what about the local situation? How will we be affected and what industries and regions are likely to be hit hardest?
Much needed research into the issue is starting to emerge, but the widely varying estimates on the job-destroying potential of AI and automation shows just how difficult it is to predict the impact of this potentially highly disruptive technology across the economy and society.
How much and how fast?
A critical question to answer is whether artificial intelligence will be as significantly disruptive to employment as previous transformative technologies, such as electricity and the internet were.
Another key question to unpick is what timeframe the transformation will occur over. Will AI and robotics rapidly supplement and replace the work of humans in the next decade across the economy, or is the change likely to happen more gradually over 30 - 40 years?
On both questions, economists have widely varying views. In May, the AI Forum, a body representing a large number of New Zealand and multinational tech companies involved in developing artificial intelligence, released Shaping A Future New Zealand, a report looking at the impact of AI in New Zealand in the coming decades.
The Ministry of Business, Innovation and Employment co-funded the research, which was undertaken by analysts and economists from IDC Research and Sapere Research Group.
They concluded that over the next 40 years, AI-driven job losses will account for just 10 per cent of overall job destruction - one million of the 10 million jobs displaced by “normal market changes”.
“This is the new electricity or the new internet,” says Ben Reid, executive director of the AI Forum, which estimates AI will contribute $54 billion to GDP (gross domestic product) by 2035.
“We've a growing body of evidence now that AI is a similarly transformational technology that will enable economic transformation within our lifetimes.”
But as transformative as it will be, he doesn’t see it leading to rapid mass unemployment. If AI evolves in a similar manner to other types of technology, which is what the AI Forum expects, mass adoption by businesses is likely to take at least 10 to 15 years.
“It was very hard before electricity to imagine how the economy would reconfigure itself. Factories took decades to move to small electric motors from steam turbines,” he says.
“There's already a lot of churn going on as tasks within jobs are automated. There'll be a longer adoption cycle than many are predicting.”
Nevertheless, the AI Forum’s estimate of job losses from automation is markedly lower than other economists’ estimates.
The automation revolution
Last week, a report by Infometrics From Education to Employment, estimated that the “automation revolution” would displace 31 per cent of jobs by 2036. That report wasn’t based on a new economic analysis, but on a review of previous reports, chiefly by PwC and NZIER and Chartered Accountants Australia and New Zealand.
NZIER/CAANZ in 2015 put the number of jobs at high risk of automation over the next 20 years at 37 per cent - 885,000 jobs. PwC, on the other hand, puts the proportion of at-risk jobs at 24 per cent in the same timeframe and notes that the figure is less than initially feared, because of our “relatively high employment rates and education and skill levels across all major demographic groups”.
So 10 per cent or 37 per cent, which one is the more likely scenario? Infometrics chief forecaster Gareth Kiernan says he struggles to accept the AI Forum’s estimate of the relatively low proportion of jobs that automation will claim.
“Yes, there is churn as people move between roles, but they are often able to continue utilising the same skill set,” he says of the millions of jobs the AI Forum suggests will be displaced by normal market forces over the next 40 years anyway.
“With automation, we’re talking about occupations largely disappearing and people being left with skills that are no longer applicable to the workforce. This process naturally happens anyway, but not at the rate suggested by the 10 million - 40-year figure.”
“My take on it is, if we’re talking about 747,000 jobs, our estimate [of jobs] at risk of automation between now and 2036, that’s almost double the rate of redundancy...over the last 17 years,” Kiernan adds.
What to do about it
Accurate estimates are important because they will inform any policy efforts in the short term to prepare for the disruption to the workforce.
Indeed, the Informetrics report dwells more on the government response in the short term to the coming labour market changes automation will bring about, rather than trying to more accurately estimate the extent of job destruction.
“In my view, the speed of transition we’re looking at over the next 20 years is significantly greater than normal,” says Kiernan.
“That being the case, I believe the government has a role to play in helping workers through that transition and trying to ensure they are able to continue contributing to the workforce and the economy.”
Infometrics suggests that lower-skilled jobs in industries like agriculture, forestry, manufacturing and the service sector will be hardest hit by automation. That means the regions could suffer disproportionately, particularly as those sectors look to automation to lower their cost of labour.
“By definition, the proposed rise in the minimum wage will increase the cost of labour provided by workers filling the lowest-skill roles – the very workers who are most at risk of being made redundant by automation,” the report notes.
The AI Forum is less concerned.
“In the context of this change, the loss of a million jobs over the next 20 to 40 years to AI does not appear to present an enormous labour market challenge,” it notes.
“Even if AI related job elimination were additional to the existing churn, it would be a relatively modest influence.”
Toby Walsh, professor of artificial intelligence at the University of New South Wales, says it is just too hard to tell what impact AI and automation will have on jobs because new game-changing applications of the technology could rapidly emerge. On the other hand, quasi-Luddite policies, such as a tax on robot labour, could be implemented to slow their take-up.
“We've no idea whether more jobs will be created than are destroyed,” argues Walsh.
“Unemployment is at record low levels, we've created lots of new jobs. But there's no guarantee that will be true going forward.”
One thing is obvious - AI will plug gaps in the workforce where it is hard to find people to do the job.
“Some industries such as healthcare, have ageing workforces, where we are already seeing shortages of workers,” says Reid.
“Automation is potentially solving some of the demographic challenges.”
With all in agreement on the disruptive potential of AI and automation, policymakers have their work cut out for themselves best preparing us to take advantage of the opportunities it presents and risks it poses.