Over the years, digitalization has led to the axing of many lower-skilled jobs, thereby polarizing the employment market. But AI could enhance some of the simpler jobs and reduce inequality.

This article by Stefan Stöcklin was originally published in UZH News on 2 October 2024. Translated by Astrid Freuler (UZH) and edited for layout purposes by the UBS Center.

Anyone who has worked in an office in recent years – regardless of their role – will have witnessed the increasing influence of computers. In the media sector for instance, the mainly female employees who used to type handwritten manuscripts into computer systems have long gone. And thanks to spreadsheet software and other office applications, secretaries now handle tasks that were formerly the remit of specialists such as bookkeepers. Another example from the finance sector are stock market traders. These days, most people carry out their market order at home on their computer, making the floor traders a relic of the past. The computerization of the work environment is a long-term trend that started with the development of powerful semiconductors in the 1970s and will continue with artificial intelligence for many years to come.

The fear of mass unemployment and the grim forecasts on which jobs will disappear next are at least as old as the rise of computers. The consulting firm McKinsey, for instance, predicts that AI will have a particularly detrimental effect on well-qualified workers with Bachelor’s and Master’s degrees and PhD conferrals. This is backed up by the services company Accenture, which claims that almost half of all hours worked in Switzerland are affected by AI.

Who gains, who loses?

The economist David Dorn, professor of globalization and employment markets at UZH, has a significantly more relaxed view of the current situation. “For a start, the consulting firms overestimate the changes brought about by AI. And they also fail to take account of the future occupations that these new technologies will enable,” says the economist. “The fear of mass unemployment is based on the erroneous assumption that there is a finite amount of work which needs to be carried out by either humans or machines. People forget that technical innovations, by dint of the new possibilities they bring, also create new fields of work that never existed before,” he explains. One only need think of all the digital information systems that were developed in countless occupational fields, thereby opening up new work opportunities. Thus even computerization didn’t lead to the great wave of lay-offs that was feared. Instead, the economies of the western industrialized nations continued to grow. Nevertheless, it is likely that the current technology drive will cause some occupations to permanently disappear.

So the question is: who will gain, who will lose? David Dorn has closely studied the impact of digitalization on the employment market. Looking back on the changes that occurred from 1990 onwards, there’s a clear pattern of mainly routine tasks being taken over by computers. Thanks to defined software programs, computers outperform humans when it comes to work that follows a set procedure. In industrial production for instance, computer-controlled robots manufacture all kinds of parts, from car doors to loudspeaker casings.

In the office environment, they have taken over all data management – storing, processing and retrieving information according to set rules. In many fields, the advantages of computers meant they superseded the workers, both in the manufacturing and service industries as well as in administration. However, machines were less suited to developing new, creative ideas and solving problems, let alone handling communication. The abstract and creative work of managers, programmers, engineers and researchers thus gained in significance. And rather than being replaced by machines, highly qualified workers benefitted from the computers, as these took care of repetitive tasks and allowed them to increase their productivity.

As a result, computers have led to a polarization in the world of employment – while routine work that could be digitalized became less important, there was a greater demand for people able to take on high-skilled roles. David Dorn and other researchers have empirically established that this pattern applies to many countries in North America, Europe and East Asia. An interesting knock-on effect can be observed in the area of so-called manual roles performed by waiting staff, chefs, chauffeurs, hairdressers and cleaning personnel. Due to the increased economic productivity, in part brought about by digitalization, there has been greater demand for their services. The losers on the job market have so far been the less qualified and lower paid employees in the mid-level segment, who performed a high quota of repetitive work that could be digitalized. Their salaries stagnated, which led to a rising inequality on the employment market.

AI broadens competencies

With artificial intelligence, digitalization has entered a new phase that is likely to result in further shifts. Where will the journey take us? David Autor, renowned economist at the MIT near Boston and long-standing co-author of David Dorn, has a surprisingly optimistic outlook. In his view, AI has the potential to upgrade the less qualified roles of the mid-level segment and reduce polarization on the job market. “AI can push back against the process started by computerization — to extend the relevance, reach and value of human expertise for a larger set of workers,” the author writes in a working paper of the NBER (National Bureau of Economic Research).

At the heart of his considerations are the subjects of expertise and specialist knowledge. While computers previously took over systematic tasks, AI can provide specialist knowledge, which could enable many workers to increase their expertise and take on more highly qualified roles.

This enhances the skill set of averagely qualified workers, allowing them to perform duties that were previously the remit of highly qualified professionals. David Autor cites operational areas such as those of doctors, lawyers, programmers and lecturers. Information technology and AI could enable nursing staff in hospitals to carry out functions which are currently the preserve of doctors, such as inserting a catheter or analyzing X-ray images. Thanks to AI, law practitioners with basic training could one day make decisions which at present only lawyers with advanced training are able to make.

It’s a similar picture in the education sector, where less qualified staff could hold lectures and carry out the work of professors. And there’s no doubt that programmers can massively increase their productivity through AI. In all these cases, writes David Autor, AI wouldn’t replace the experts, whose qualifications will still be in high demand, but will provide less qualified workers with a certain degree of expertise. This particularly applies to the healthcare system and the software industry, where the demand for expert knowledge is expected to rise markedly.

What is the value of specialist knowledge?

Only the future can tell how realistic this somewhat optimistic scenario is. David Autor does concede that AI could devalue the as yet highly-prized expertise of some professionals, in the same way that navigation systems have made the extensive geographical knowledge of taxi drivers obsolete. More commonly though, innovation creates new possibilities and, with them, also new expert roles. According to Autor, this makes it all the more important to take a measured approach when introducing the new technology of artificial intelligence and to do so in collaboration with those affected and the relevant societal bodies. For Autor, AI offers an opportunity to halt the polarization of the job market brought about by computerization during recent years.

David Dorn feels the scenarios his colleague presents for some occupational fields are plausible, but takes a more cautious stance. For him, the question of where the AI journey will lead remains open, as there is no relevant employment market data. Dorn expects the current trends to continue in the immediate future, meaning there will be further automation of certain functions in the services and production sectors.