In the first policybrief, the ai:conomics research team has compiled the current state of knowledge on how AI impacts work tasks and skill profiles, and how its implementation affects the well-being of workers in the workplace. They also identified what challenges researchers face in studying AI in the work context and provided a foresight on what it will take to expand the body of knowledge and what added value that has for shaping the world of work in the future.
The potential of AI vis-à-vis previous technologies
AI can be seen as a general purpose technology (Lane and Saint-Martin, 2021) . This distinguishes AI significantly from preceding technological innovations. More specifically, the technology does not only depend on instructions or rules provided by humans, but also consists of self-learning; AI can automatically infer connections between inputs and outputs without the need for any manual rule-based design made by humans. On the one hand, AI thus offers new opportunities to support humans in tasks that require analytical thinking. For example, in medical diagnoses or prognoses. On the other hand, the technology can also replace human skills. This is possible for tasks that can be codified, for example in speech and image recognition.
Researchers estimate the impact of AI in the world of work to be much more far-reaching than that of previous technologies: To a much greater extent, AI can change the task structure of jobs by both replacing some activities while simultaneously creating novel ones that often require new skills.
The impact of AI on work and workers
Will AI replace tasks or create new ones?
Science distinguishes between complementing and substituting impacts of artificial intelligence (AI) in the work context.
Complementing, or supportive, AI could lead to significant productivity gains and support human workers in such a way that AI use leads to increased employment in the long run. The literature suggests that complementing AI already exists. For example, it is already being used in oncology as a decision support tool in cancer detection. In addition to supporting existing occupations, AI can also create entirely new occupations that do not yet exist. In the example given, this would apply to radiologists, who develop AI applications in the technical innovation phase. Furthermore, AI needs excessive training based on real-life data of good quality. This data often needs to be created first and therefore creates new job opportunities for AI- and other types of experts.
In contrast to complementary AI, substitute AI focuses largely on automation, replacing human skills. This is already being applied, for example, in the field of speech and image recognition and natural language processing. AI applied in this way is likely to have more of a dampening effect on demand for workers in occupations that can be automated. Researchers suspect that rising productivity growth cannot compensate for the reduced need for human-performed tasks in the work process in which AI is used. They therefore expect different effects of AI use in jobs that can and cannot be automated.
As it appears, both effects of AI - substituting as well as complementing - may occur simultaneously in the labour market. This makes it difficult to assess the pure overall impact of AI adoption on the future of work. A future with AI will depend largely on the type of AI deployed, as well as the nature of the activities and sectors in which it is deployed.
What skills are in demand in the AI work context?
Findings from a recent OECD 2021 paper indicate that social-emotional skills will become increasingly important in addition to the pure technical skills needed to develop an AI. This is because AI has many capabilities. Nevertheless, at this stage AI cannot do everything at yet. Especially in tasks that require a higher level of interpersonal and social skills, humans have a clear advantage. These skills are needed, for example, to work in interdisciplinary teams, be able to communicate and present existing results and to creatively solve problems. But they are also in demand in the area of knowledge transfer, for example in the area of continuing education. These skills thus contribute to ensuring not only that all stakeholders understand AI, but also that it is used correctly.
To date, there has been little empirical research analyzing the detailed skill needs in response to the increasing use of AI. Understanding how AI will change the way we work, and thus what skills we will need to do our jobs well, will be critical.
AI in the workplace: How might the use of AI affect employees?
So far, there is little evidence in science about how AI will affect the nature of work. Instead, several studies have looked at how other automating technologies have changed the work environment over the past decade. In the policy brief, researchers part of ai:conomics summarized an overview. In addition, they have scrutinized the small number of studies that examines but do not yet empirically assess the impact of AI on workforce well-being. They find that well-designed and implemented AI could have a positive impact on workers by encouraging them to be more autonomous, flexible, and creative. Some studies, however, suggest that AI-influenced changes in the workplace can lead to a decline in mental health and job satisfaction.
All in all, the state of research suggests that the introduction of AI presents both significant opportunities and challenges for workers who must adapt to workplace changes and the work environment of the future. To date, however, there is limited research on how AI affects workers' well-being.
Investigate AI where it is being used
More firm-level data is needed
Crucially, we recognize that while profound change is inevitable, AI-related practices will take different forms for each company and individual in the labour market. As all companies deploy different types of AI, depending on their long-term goals, strategies, and expectations, the underlying mechanisms are likely to impact their work and the people who perform it differently. With their individual deployment and use in diverse industries, the potential consequences of AI can therefore be better understood by analyzing company-specific, customized AI applications. Research of this type, however, is currently scarce. Such detailed firm-level analyses can provide greater clarity in determining the role of AI in the workplace.
With more firm-level insights and multi-stakeholder involvement, decision makers from businesses, governments, as well as social partners can use this as a basis to develop and promote measures that will enable society to reap the benefits of AI while mitigating its potential risks.