By Mark Fairlie
Despite growing fears in recent years about artificial intelligence replacing the human workforce, new research from the Organisation for Economic Cooperation and Development (OECD) suggests the threat may not be as serious as many had previously suspected.
Lowering earlier forecasts
Back in 2013, a report from Oxford University titled ‘The Future of Employment: How Susceptible Are Jobs to Computerisation?’ predicted that almost half of all jobs in the USA and 35% of jobs in the UK were at “high risk” of being automated by 2033.
The paper has been extremely influential in discussions around the threat of AI; forming the basis for many different projections from various government bodies, inspiring other studies with similar findings, and even influencing the BBC’s popular risk-prediction tool for workers to see how vulnerable their sector may be to automation.
Despite the findings from this research, new analysis from the OECD has cut these predictions down by more than half. According to the latest report, around just 10% of jobs in the US and 12% of jobs in the UK are considered at high risk of being replaced by AI.
The latest findings
In addition to these lowered percentages, the OECD report also offers a more in-depth analysis as to the types of jobs most likely to be replaced by artificial intelligence. The organisation noted that previous forecasts had greatly exaggerated the impact of automation since they had relied on the broad grouping together of jobs with the same title.
The new analysis takes into account the differences between these jobs, such as how the role of an engineer might vary greatly depending on the project or employer, meaning some workers could be more vulnerable to automation than others with the same title.
The OECD noted that there are a wide variety of factors which could make similarly-titled roles more or less susceptible to automation, such as whether the role requires a human touch in the form of creativity, complex reasoning, or developing social relationships.
It was found that the risk of automation declines with the level of education a worker has achieved, their level of measured skill such as numeracy and literacy, and their wage level. The OECD concluded that this suggested automation is very much “skill-biased”, with AI affecting low-skilled jobs more significantly than any previous waves of automation.
The study also revealed that the risk of automation is much higher for jobs by teenagers, with the highest automatability being found among jobs traditionally held by young people. It suggested that young people could find it harder to find employment since entry-level posts including retail and personal care were at much higher risk of automation than roles requiring more experience.
“Youth and adults do different things at work, even when they hold jobs with the same occupational title,” the report said. “The warnings in some developed countries that teen jobs have been harder to come by in recent years should be taken seriously and studied in the context of job automation.”
Global impact of automation
The OECD analysis used data collected in a survey of thirty-two member nations in order to create a more comprehensive image of today’s workforce and the threat of automation across the globe.
Overall, the organisation discovered that 14% of jobs across the nations surveyed were found to be at high risk of being taken over by AI – equating to more than 66 million jobs worldwide. They also noted that an additional 32% of workers will experience significant changes in their job content as a result of increased automation in the near future.
The study found that the risk of automation varies greatly from country to country, with jobs in Anglo-Saxon and Nordic countries being considerably less automatable than those in Eastern European countries, South European countries, Germany, Chile and Japan.
The OECD identified Britain as one of the countries that will be least affected by automation, however, they noted that the level of skill a job requires was the main deciding factor in determining the real risk of automation.