This page is approximately a 5 minute read
This page was published on
Professor Milena Nikolova, University of Groningen
Professor of the Economics of Well-being and a Rosalind Franklin Fellow at the University of Groningen. Between 2014 and 2017, Dr. Nikolova was a Research Associate at the Institute of Labor Economics (IZA) in Bonn, Germany, where she is now a Research Fellow (since January 2018). She is also a Nonresident Fellow at the Brookings Institution and Bruegel, and a Fellow at GLO.
Workplace harm is a major global safety challenge. The 2024 World Risk Poll shows that 18% of the global workforce – around 667 million people – experienced serious harm at work in the past two years, with risks concentrated in certain industries and among workers with limited protection and training. At the same time, robots, AI, and digital technologies are reshaping how work is organised, altering both workers’ exposure to physical risks and the ways risks are monitored, reported, and managed.
Evidence shows that some technologies, such as industrial robots, can lower the physical burden of tasks and injury risks. However, technological change may also affect safety in less visible ways, influencing mental health, trust in safety systems, and whether workers report harm. We lack a clear global understanding of how these technologies interact with worker characteristics, institutional quality and safety practices to shape both actual workplace harm and workers’ experiences of risk. This project addresses that gap by analysing how technological change influences workplace harm, safety, and wellbeing across countries and institutional contexts.
There are two parts to this project. The first is a global analysis of individual-level data from the World Risk Poll, covering more than 140 countries. This will link workers’ reported experiences of workplace harm to country-level measures of technological change, including industrial robots, information and communication technologies, and artificial intelligence readiness. This analysis also accounts for industry of employment, access to safety training, worker protections, and institutional quality. It will enable identification of global patterns of vulnerability and understanding of how technological change relates to workplace risk across different contexts.
Secondly, there will be a controlled survey experiment in the Netherlands using a nationally representative panel. Participants will evaluate realistic workplace scenarios with different safety systems – human-led, AI-led, and hybrid. By randomly assigning these scenarios, the experiment will help identify the causal effect of safety systems on perceived safety, stress, trust, and wellbeing within a specific context.
This project aims to improve safety for workers in sectors and settings where exposure to workplace harm remains high and access to effective safety systems is uneven. The data from the 2024 World Risk Poll shows that workplace harm is highest in specific industries, such as fishing, construction, mining, and agriculture, and among workers with limited access to occupational safety and health training. These patterns appear across countries and income levels, although they are more pronounced in lower-income settings.
This project aims to improve safety for these groups by clarifying how technological change interacts with industry context, worker characteristics, and institutional settings to shape both experienced harm and reporting behaviour. By showing where technology coincides with lower levels of harm and where it is associated with under-reporting, stress, or reduced trust, the findings will help policymakers and employers target safety interventions more effectively. In addition, experimental evidence on different workplace safety systems can clarify how workers respond to human-led, AI-led, and hybrid approaches. This will support the design of safety systems that workers trust, engage with, and use to report risks before serious harm occurs.
The World Risk Poll enables this project by providing globally comparable, individual-level data on workers’ own experiences of workplace harm, safety concerns and experiences, and access to occupational safety and health training, collected through nationally representative surveys in over 140 countries. Unlike administrative injury statistics, which rely on formal reporting systems and vary widely in coverage and quality, the Poll captures self-reported harm directly from workers themselves. This includes groups that may be missing from official data, such as informal, self-employed, and precariously employed workers.
In addition, the dataset contains socio-demographic information on respondents. This makes it possible to analyse how workplace harm and safety perceptions vary across worker characteristics using a consistent methodology.
I want to engage with researchers, policymakers, and organisations working at the intersection of work, safety, and technological change. This includes academic audiences through international research networks, including the Global Labor Organization and the International Society for Quality of Life Studies, to ensure the findings inform and connect to ongoing scientific debates.
I also want to engage with policymakers and public bodies involved in occupational safety and labour market governance, such as the ILO, OECD, and national institutions such as the Employee Insurance Agency (known as the UWV) in the Netherlands. Engagement with employers’ organisations, such as the Confederation of Netherlands Industry and Employers (known as VNO-NCW) in the Netherlands, is important to understand how safety technologies operate in practice.
Finally, I aim to connect with practitioner organisations and other World Risk Poll grantees, such as the Royal Society for the Prevention of Accidents, whose work on accident prevention, training, and advocacy provides a direct pathway to translating evidence into practical safety improvements.
The Global Network of Civil Society Organisations for Disaster Reduction (GNDR) will use World Risk Poll data to help identify those most endangered by severe weather and aim to improve their access to early warnings.
The Resilient Cities Network are partnering with Arup to use World Risk Poll data to enhance how cities assess environmental, social and economic risks, improve decision making and target investments.
Dr. Sarah Jenkins and Dr. Andrea Taylor discuss the integration of the Poll with data on vulnerabilities, governance, and long-term climate projections to identify predictors of disaster risk perception and preparedness, and develop early warning and risk communication strategies.
