Artificial Intelligence (AI) is no longer an abstract vision of the future but an integral part of everyday business reality. It is transforming the tasks people perform at work, how work is organized, and how decisions are made. The rapid spread of generative AI has accelerated this development in recent years. This raises important questions:
How is AI already changing work in companies today? What actions can companies and policymakers take in response? Which mechanisms are at play?
This policy brief addresses these questions: it provides an overview of the current state of research on the topic by synthesizing findings from field studies at the company level and presenting them in an accessible way. The view extends beyond the ai:conomics project: in addition to the results of ai:conomics, insights from the broader research literature are also presented. Furthermore, concrete recommendations for action are formulated on how the introduction of AI can be consciously shaped both politically and at the corporate level to avoid or mitigate worst-case risks.
In this way, a compact insight into the current knowledge on the impact of AI on the world of work is provided, which can serve as a basis for informed decisions for decision-makers in companies, politics, and social partnerships.
Key Insights:
- AI is no longer an abstract vision but an integral part of business reality, particularly through the rapid spread of generative AI technologies. It is changing tasks, how work is organised, and decision-making processes across industries.
- Unlike previous automation technologies, AI particularly affects highly skilled occupations and expert tasks. This creates fundamentally different distributional effects with broad implications for labour market policy.
- AI can increase productivity and work quality, especially for less experienced workers, but its effects are highly context dependent.
- AI is transforming work primarily through shifts in tasks and skill requirements rather than mass job losses. However, agentic AI systems may significantly increase displacement risks in the future.
- Successful AI implementation depends less on technology and more on governance, work design, communication, trust, and participation. Middle management, domain experts, HR, and works councils play crucial role in achieving positive outcomes.
- Companies, social partners, and policymakers are called on to actively shape this transition so that productivity gains are widely shared and risks to workers are addressed early.
How does AI deployment affect productivity – and what factors determine this?
AI makes even highly skilled tasks automatable, impacting professions that were hardly affected by previous waves of automation. It can improve efficiency and work quality, for example by supporting on-the-job learning, especially for less experienced workers. At the same time, the potential applications are unevenly distributed across employee groups and are primarily evident in highly skilled knowledge work. However, without the right framework, performance declines can also occur: AI is not a plug-and-play general-purpose upgrade, but requires targeted work design and supportive organizational structures. Direct AI-related job losses are so far only limited in evidence (Acemoglu & Restrepo, 2019; Özgül et al., 2024), but with increasing adoption, they may become more significant, though likely not equally for all employee groups.
How does AI deployment affect employees?
The most significant current impact of AI on employees is the redesign and restructuring of work. Tasks are increasingly shifting toward the monitoring and control of AI systems, dealing with exceptions, and decision-making. This can promote autonomy, learning, and competence development – but it can also increase stress, surveillance, and the loss of tacit experiential knowledge. Employees often prefer the new tasks supported by AI but tend to favor AI as support in their familiar work rather than additional, more complex tasks that remain when AI acts autonomously. Overall, the effects on well-being are ambivalent: the technology itself is less decisive than its design – task and process design, leadership, and the organizational context. AI deployment is therefore always a matter of workplace design and handling psychosocial risks.
How does AI adoption impact skills?
AI changes how explicit and implicit expertise is used: it increasingly automates expert-level work, thus restructuring the architecture of professional skills. Current research suggests that substitution is the most common consequence: tasks based mainly on explicit knowledge are fully automated and disappear. In contrast, the demand for AI-related skills rises. While AI can take over cognitively demanding tasks, allowing people to only review the results, this enables less experienced workers to take on more complex tasks. Nevertheless, widespread de-skilling has not yet occurred, as many tasks still require tacit knowledge and therefore differentiated judgment and expertise.
Agent-based AI as the next stage of automation: What impact?
Agent-based AI systems can not only generate content but also independently execute entire task packages, make decisions, and act autonomously within operations. This represents the next level of automation. So far, there is limited scientific insight into their impact on employees, businesses, and the economy. Three key risks are identified: technical capabilities are advancing faster than accepted limits of use, skill requirements are further shifting, and highly skilled tasks are increasingly decoupled from location, meaning anything that can be standardized or broken into workflows is becoming globally tradable.
