By Dr. Volker Stocker, Head of Research Group at Weizenbaum-Institute and Postdoctoral Researcher at TU Berlin.
On September 11th, we organized an exclusive policy panel that brought together four experts, each bringing complementary perspectives and insights, to discuss AI and the Future of Work. We set out to explore organizational and societal shifts fueled by AI based on the evidence of how jobs and skills are changing, and aimed to critically reflect on the adequacy of Europe’s regulatory and normative responses. Above all, we aimed to discuss how we can ensure a human-centred AI transition. This essay offers a recap and key insights that emerged from the panel.
Introduction — (Generative) AI as a Change Agent
Artificial Intelligence (AI) has become an increasingly pervasive force, widely expected to profoundly shape and reshape almost every aspect of our increasingly digital economies and societies. Recent developments surrounding generative AI (GenAI) have further fueled these expectations, inspiring visions of digital futures shaped by an expanded range of human-machine collaborations and unprecedented opportunities for task and skill automation.
AI is often portrayed as a transformative catalyst or change agent, one that will fundamentally and irreversibly alter how humans gather information and learn, how they interact and collaborate (with machines), and how work is organized. Such visions feed not only optimism about productivity gains and growing opportunities for automation and human augmentation. They also spark concerns about job security, shifting skill and job demands, more worker surveillance, and the adequacy of existing policy and regulatory frameworks.
Beyond the hype, fundamental questions emerge: How can AI be integrated in the workplace — by firms and workers — and what does this mean for the future role of humans in the work loop? How will AI innovation shape workplace realities and working conditions? Which jobs and skills will (continue to be) in demand, and which may disappear? And, crucially, are our regulatory frameworks fit for the challenges ahead?
Insight #1 — Labor Market Effects and Impact on Workers are Real but Uneven
The pace of AI adoption is accelerating, with estimates suggesting that AI could affect 70% of job skills by 2030. However, the implications for job replacement, upskilling, and reskilling vary considerably across different occupations, sectors, genders, and company sizes. Mirek Pospíšil (LinkedIn) shared nuanced empirical insights into how AI is reshaping jobs and skills from the vantage point of LinkedIn, distinguishing between three groups of occupations that are affected by AI in different ways — each representing roughly one-third of all jobs. The first group includes occupations that are most susceptible to being disrupted by AI, such as legal associates or customer service representatives. These roles face higher automation risks as skills are increasingly replicable by AI, making upskilling, reskilling, and targeted transition support critical. Recent evidence suggests that women and younger workers are among the groups particularly exposed to such AI-driven disruptions. The second group encompasses occupations that can be augmented by AI, such as customer experience managers or food consultants. These roles involve tasks that are partially replicable but heavily dependent on (and thus complementary to) human input. The third group encompasses occupations that are relatively insulated and not easily automated or replicated by AI, such as construction specialists and roughly one-third of leadership roles.
These insights — and the underlying evidence of AI’s uneven labor market effects and their impacts on workers — are supported by worker surveys such as Eurofound’s Working Conditions Survey (EWCS). According to the EWCS, only 12 % of workers across the EU-27 countries use GenAI tools in their work, with numbers ranging from 25% in Luxembourg to 5% or less in Greece, Italy, Portugal, Romania, Serbia, and Bosnia-Herzegovina. The survey also reveals notable disparities and adoption gaps across genders and age groups, with men and younger workers more likely to use GenAI tools. Similar differences are observed in the use of various algorithmic management approaches, including the automated allocation of work tasks, monitoring of work performance, and automated work scheduling. Adoption rates are highest in financial services and transportation, and lowest in construction and agriculture. Drawing on these insights, Anna Ludwinek (Eurofound) highlighted that AI-driven task automation is shaping the evolution of the labor market and influencing several dimensions of job quality, including work intensity, skills and discretion, career prospects, and earnings, as captured by the EWCS.
Insight #2 — Evidence-based Policies Are Needed to Guide the EU Way Forward
The inherent uncertainties associated with the tremendous technological dynamism related to AI — and resulting limitations in our ability to forecast our digital future(s) — must be acknowledged when designing adequate and effective policy responses. Mirek Pospíšil emphasized that current empirical insights suggest that one-size-fits-all policy responses would be inconsistent with evidence-based policymaking. This view was echoed by Max Uebe (European Commission). Beyond highlighting the importance of building trust in embracing and adopting AI, from both the perspective of workers and companies, he emphasized that the benefits of AI, such as automating repetitive tasks, must be carefully balanced against potential risks and adverse effects related to algorithmic management, increased surveillance, AI bias and discrimination, and data breaches. Addressing this delicate balance requires forward-looking, probably adaptive policy responses. In this context, Max pointed to the Quality Jobs Act, which was announced by Commission President von der Leyen in her recent State of the Union address and aims “[t]o ensure that modern employment keeps pace with the modern economy.”
Insight #3 — There Is More to AI Regulation than Products: Towards AI Policies Grounded in Human-centered Perspectives
AI policies should be grounded in human-centered perspectives. This was emphasized by Humbert De Biolley (Council of Europe), who explained that workers bring more than just labor to work — they bring their individual (human) rights. Accordingly, product-centric AI policies such as the AI Act should be complemented by frameworks that approach AI through a human rights lens, focusing on individuals to facilitate a genuinely human-centered AI transition. The Council of Europe’s Framework Convention on AI, signed by 17 parties including the EU and the USA, embodies such an approach and incorporates a human rights lens. To ensure compliance with the fundamental principles enshrined in the Convention — including human dignity and individual autonomy, equality and non-discrimination, and transparency and oversight — the Council of Europe is developing a methodology for assessing risks of non-compliance via context-based risk evaluations. Supporting this perspective, Anna Ludwinek emphasized the need to shift from a productivity-centered to a human-centered perspective, highlighting the importance of worker participation in facilitating a human-centered AI transition. This may require recalibrations to strengthen social dialogue, data rights, and workplace practices.
Concluding Remarks based on the Panel Discussion
AI is more than just another typical technological wave. It represents an evolving range of technologies that are profoundly transforming how we work, how organizations operate, and how we understand rights, fairness, and individual dignity. While AI offers unprecedented opportunities, it also brings risks and uncertainties — particularly regarding the future of jobs, skill demands, and worker vulnerabilities. Striking the right balance between these opportunities and risks will be critical.
To shape a human-centered, AI-driven future of work, we need inclusive and sustained dialogue with contributions from multiple stakeholders — including policymakers, researchers, companies, social partners, and workers themselves. This way, we can hope to design policies that ensure AI serves people — and not the other way around.