Artificial intelligence and robots: which jobs are most likely to replace them

  • According to research, the next wave of automation will affect the entire world of work.
  • Some jobs are more at risk than others.
  • But the risk of being replaced by robots can be reduced.

More digital topics can be found here

What distinguishes physicists from meat packers? The researchers, led by Antonio Paolillo of the École Polytechnique Fédérale in Lausanne, Switzerland, believe the risk of being robbed by artificial intelligence (AI) or robots is particularly important. They calculated the “risk of automation” for nearly 1,000 jobs and published it in the journal Science Robotics. Physicists have the lowest risk of losing their jobs to the machine, while butchers and meat packers have the greatest risk.

The general conclusion of the researchers is both pessimistic and optimistic. Many jobs are mostly human skills that AI and robots (or a combination of them) can also perform. Scientists have found few jobs where less than half of the required skills can be performed by machines. Machines could therefore take over many activities and replace some of the labor force in most jobs.

On the other hand, scientists write that, with relatively little training, you can often find a similar job for which the risk of automation is much lower. There is therefore a risk of mass unemployment that can be avoided with a skilful training policy.

The robots leave the cages

Until now, robots in particular have performed low-skilled, physical and monotonous work, for example in the automotive industry. There are three million robots in industrial production around the world, three times more than at the beginning of the 21st century. Industrial robots usually work in safety cages due to their rigid and programmed movements.

Thanks to advances in artificial intelligence and robotics, robots also learn more complex jobs that require manual dexterity, flexibility and collaboration with humans. Thanks to lightweight grippers made of plastic and controlled by compressed air, robots are becoming more sensitive and safer for the environment – that is, they can leave the cages. They handle sensitive goods and, thanks to artificial intelligence, free themselves from rigid, programmed sequences of movements. Prototypes even collect sensitive fruits such as raspberries, and artificial intelligence detects, among other things, the ripeness of the fruit.

Artificial intelligence makes robots versatile, so they can do a lot of work, for example in slaughterhouses. The EU research project RoBucher is developing an AI-controlled robot designed to cut and boning animal carcasses in a process that has so far involved humans.

These changes are trends. Almost the entire food industry is now asking about automation, Dieter Rothenfusser, portfolio manager at Kuka, a robot manufacturer, said in a recent interview.

As robots cope better in the human environment, they will also be used more and more often as assistants, e.g. to relieve nursing staff from heavy physical work. But robots can also partially meet the emotional needs of those in need of care, as demonstrated by the humanoid robot “Pepper”, which recognizes and responds to emotions.

Comparison of human skills with the skills of machines

Intellectual work is also not safe from automation. AI often outperforms human experts such as skin cancer diagnosis. Journalists and authors will soon hand over a number of routine tasks to AI, such as writing newspaper articles. In any case, so-called “language models” formulate texts that are read as if they were written by humans.

Swiss researchers took note of the progress. “We assess the likely impact of the next wave of robotization on nearly 1,000 professions,” they write. They first analyzed what human skills are required for each of these tasks. They used the US Department of Labor’s online O * Net database, which includes job descriptions.

Then they compared the catalog of individual activities with the skills that robots may have. They took this from the ‘Multi-year Robotics Roadmap’ published by SPARC, a public-private partnership between the European Commission and the European robotics industry.

From this data, scientists determined the percentage of skills in each job that the robot could adopt. It also covered how important the right skills are at work and how the worker needs to be trained. Some human talents appear almost naturally, like mathematical reasoning, others appear less natural, like learning by observation.

Researchers also took into account the maturity of the robots. It is measured numerically, the so-called The “technological readiness level” that the Swiss team included in its formula.

Exchange risk indicator

They sum it all up in what they call the “Automation Risk Index” or ARI for short. “ARI can be interpreted as the proportion of human skills at work that can also be performed by machines,” writes the team.

The list is surprising. Even the safest physics robot has an ARI of 0.43. So 43 percent of the activities of physicists can be automated. In fact, it’s already happening: Artificial intelligence, for example, is combing the complex particle traces that large particle accelerators leave behind in detectors such as the Cern nuclear research center near Geneva.

Inside, there are highly qualified positions such as marine engineer, assistant surgeon or singer. After all, no task can be automated one hundred percent. Butchers and meat packers face a 78 percent risk of automation. The risk of self-replacement by robots can be checked on the website set up by the researchers.

What does that mean now? Scientists write that automation drives structural change. They note that when the United States produced automated routine jobs in the late 1970s, workers shifted to lower-paying service jobs. Using their data, they examined retrospectively how workplaces at high risk of automation had developed over the past two decades. Employment in these occupations has fallen and wages have increased less than in other occupations.

The team writes that it is worth investing in retraining and further training. They estimated how complicated it was by comparing job profiles in different occupations. “We found the cost of retraining to be low more than high,” they write. They cite electrical engineers as an example. They have an automation risk of 61 percent. With little effort, they could qualify as software quality assurance technicians with an ARI of 58%. However, the researchers also explain that such measures can only reduce, not prevent, job losses caused by robots.

Is the current automation something new?

The study by Paolillo’s team is not the first of its kind. As early as 2013, Carl Frey and Michael Osborne of the University of Oxford sparked heated debates, predicting that “computerization” would jeopardize 47 percent of jobs in the United States.

Katharina Dengler and Britta Matthes from the Institute for Labor Market and Occupations Research presented similarly alarming figures for Germany in 2018. Depending on the qualifications, from 24% (for expert work) to 58% (for unskilled work) activities may be performed by computers or robots.

But not all experts fear mass unemployment as a result. They argue as follows: Earlier waves of automation initially cost jobs, especially the low-skilled. But as the machines increased the efficiency, the economy grew. This, in turn, created new, labor-intensive jobs. The next level of computerization would be exactly the same. Not everything that can be automated will be automated, add Dengler and Matthes, because “there are economic, ethical or legal aspects, for example.”

There are counter-arguments to this view. For example, this automation was slower than it is today and that people had to adapt over a generation. Moreover, digitization is also destroying highly skilled and creative jobs, making it difficult for many victims to acquire even higher qualifications. After all, greater productivity does not necessarily lead to economic growth, as many markets are already saturated, to put it simply: consumers who already own two cars do not need a third.

gaps in the study

Among the studies to date on this subject, the most promising is the one conducted by Paollilo and colleagues – comments Andrea Gentili from the University of International Studies in Rome. However, the economist emphasizes that the risk of losing work to the machine is very difficult to estimate. More accurate data is needed to make the forecasts more reliable. For example, O * Net job profiles are not complete.

Nevertheless, the present study shows one thing in the first place: the next wave of automation will affect the entire world of work. AI and robots will revolutionize the economy. Paolillo’s team sees his method as helping politicians to cope with structural change: “Governments could use the proposed method to assess the risk of unemployment among their populations and adjust education policy.”

Sources used:

  • Science Robotics: How to Compete with Robots by Assessing Job Automation Risks and Resilient Alternatives
  • International Robotics Federation (IFR): World Robotics 2021
  • Robots in the food industry: “Almost the entire industry asks for automation”
  • SPARC: 2020 Multiannual Robotics Action Plan
  • Algorithms as particle discoverers
  • University of Oxford: The Future of Employment: How Computerized Workplaces Are?
  • Labor Market Institute and Vocational Research: Few job descriptions keep up with digitization
  • Science Robotis: Answering a great question about automation
This post is from the RiffReporter journalism portal. On, around 100 independent journalists report together on current affairs and general information. RiffReporter received the Grimme Online Award for its offer.
ionized radiation

You can read the radiation exposure in your area on the map of the Federal Office for Radiological Protection. Map available at: (photo: iStock-shcherbak volodymyr)

Leave a Comment