A working paper written for the Organization for Economic Cooperation and Development estimates that about 14 percent of the jobs in 32 OECD countries, including the U.S., are at high risk of being automated.
That raw figure may not sound as dire as some of the previous numbers cited for the effect of automation and artificial intelligence on employment, and that’s what’s been grabbing the headlines over the past couple of days. But a close reading of the report, published last month, shouldn’t lead anyone to brush off the issue — as Treasury Secretary Steven Mnuchin did last year.
The authors of the study, Ljubica Nedelkoska and Glenda Quintini, say the level of automation risk varies widely from country to country. Slovakia comes in on the high side (33 percent), while the projected risk is only 6 percent in Norway.
The high-risk percentage for the U.S. is 10 percent, which is significantly lower than the 47 percent that was cited in a provocative 2013 study by Oxford researchers. But even 10 percent translates to about 15 million U.S. jobs.
“This would amount to several times the disruption in local economies caused by the 1950s decline of the car industry in Detroit where changes in technology and increased automation, among other factors, caused massive job losses,” the OECD study’s authors point out.
What’s more, automation and AI will introduce significant changes in the nature of work for far more jobs than those at high risk of being fully automated. Close to one in two jobs are likely to be significantly affected, the authors write. It’s just that the effect will be highly variable.
In other words, to paraphrase science-fiction author William Gibson, the future won’t be evenly distributed.
That perspective is relatively in sync with other recent forecasts, including a study issued last November by McKinsey Global Institute. That report said as many as 38.6 million Americans may have to switch occupations or learn new skills to hold down a job in 2030.
Nedelkoska and Quintini argue that their findings are truer to reality than the 2013 study because they were able to capture more of the variability within occupations. The earlier study took 70 data points into account, while the study conducted for the OECD reflects 4,656 individual observations of job skills.
Among other findings from the OECD study:
- Generally speaking, jobs in Anglo-Saxon and Nordic countries, plus the Netherlands, are less automatable than jobs in Eastern and Southern European countries, Germany, Chile and Japan.
- Automation is projected to have the biggest impact on the manufacturing industry and agriculture, plus a number of service sectors. Low-skilled jobs were seen as most at risk, and the authors rejected the view that automation and AI could soon have an adverse effect on highly skilled occupations.
- Some low-skilled occupations — for example, personal care — are unlikely to be significantly affected.
- The risk of automation was seen to be highest for jobs traditionally filled by teenagers. “Automation is much more likely to result in youth unemployment, than in early retirements,” the authors say.
The authors said their findings highlighted the need for a greater focus on internships and worker retraining. The OECD stressed that the report reflected the authors’ conclusions rather than policy recommendations by the organization or its member states.