An online index of nearly a thousand jobs may useful in cluing in folks to the automation risk their field of employment faces.
At the bottom of the list – meaning they’re the most likely to be replaced – are meat packers and slaughterhouse workers, while the least likely to see their jobs automated are physicists. In general, those in food services, grounds and maintenance, and the construction industries are most at risk. Those in education, social services, and management least so – but not entirely.
Scientists at École Polytechnique Fédérale de Lausanne and the University of Lausanne in Switzerland collaborated to draw up the index as they contemplated the potential social impacts of automation. While their list is targeted at individuals, the team said there’s other uses for the methodology they developed to create the automation risk scores featured in the index and map career transitions for displaced workers.
In particular, a paper the team wrote about their work argues that “governments could use the [underlying methodology] to evaluate the unemployment risk of their populations and to adjust educational policies,” and “robotics companies could use it as a tool to better understand market needs.”
The researchers compiled a list of jobs from the Occupational Information Network – a public dataset of 967 jobs and their relevant knowledge, required skills, and necessary abilities. The knowledge, skills, and abilities (KSAs) for each job were compared to the likelihood that an AI or robot could do them, with specific robotic tasks defined by SPARC’s Robotics Multi-Annual Roadmap.
To add to the analysis process, the likelihood of an AI or robot taking said job was weighed using the EU’s tech readiness levels [PDF], which assess how close to realization new technologies are.
The researchers didn’t only want to learn who was at risk, they also wanted to create a method for comparing jobs to suggest more resilient careers for those at risk of automation job loss.
To figure that out, a two-part formula was created with two results: an Automation Risk Index (ARI), and Resilience Index (RI).
For the former, human KSAs are compared to relevant robot and AI abilities to give a number between 0.43 (low) and 0.78 (high) automation risk.
Take news reporters and correspondents, for example: our ARI is 0.58, just a little less than half way up the scale. Computer programmers are a tick higher at 0.59, while network and systems administrators sit just a bit lower at 0.57.
Those in the high risk category include the entirety of the combined food prep and serving industries – which have an ARI of 0.72 – those working in the mining industry (0.69) and “helper” jobs in construction and service industries.
The lowest scores can be found among postsecondary education (low 0.5s), physicians and other medical careers, and the legal profession.
The RI portion of the equation generates a number – the lower the better – that indicates “job transitions with better trade-offs between automation risk and retraining effort,” the paper explains. Jobs with relatable skills, at least according to the way the team built its model, are sorted by ARI score and the amount of effort required to learn relevant KSAs.
The online tool includes three career alternatives for each of the 967 occupations researchers looked at. Reporters and correspondents, the tool says, ought to consider becoming post-secondary law teachers, physicists, or mathematicians.
That’s the exact same advice it gives to actors, poets, lyricists, and creative writers, and nearly the same advice it gives to musicians (just swap law teacher for statistician). These professions may seem at first blush unlikely to be replaced by robots any time in the near future. Nonetheless it’s good to know that musicians can fall back on statistics if necessary. Dylan, for example, could plot a chart showing exactly how many roads a man must walk down. ®