Conference Agenda

As the first wave of lockdowns swept the globe, the least fortunate workers simply lost their jobs. Of those who kept them, many ‘knowledge workers’ were permitted to work from home, while many ‘essential workers’—e.g., in healthcare, food production, and public services—had to continue working in person. Both groups, however, experienced a rapid uptake in fine-grained monitoring and ‘algorithmic management’ technologies. Employers installed software to monitor remote workers’ keystrokes and mouse movements, take screenshots of their screens, and take photos through their webcams. Some of these systems were so disruptive that a market in ‘mouse mover’ devices developed to let remote workers leave their computers to take bathroom breaks. In-person employers also intensified surveillance, deploying finer-grained worker location-tracking and work quotas—in some cases even firing workers algorithmically, with no intervention by human managers. In severe cases, workers reported being unable to take bathroom breaks because of algorithmically-increasing quotas. Algorithmic systems have also been deployed in hiring: industry research suggests over 95% of the Fortune 500 have adopted automated systems that rank applicants by scanning their CVs; some companies have also adopted machine learning-based video interview software. The severe efficiency pressures on employers make it easy to understand the motivations for adopting these technologies. But they also pose severe—and largely underappreciated—risks.

We report the findings of a two-year interdisciplinary review of literature on algorithmic management in economics, policy, and law, including recent investigative journalism and legislative developments. We report three main findings. First, algorithmic management poses new and significant risks to workers’ fundamental rights, to decent work, and to labour market institutions. We identify the mechanisms producing these risks: increased privacy harms; widening of information asymmetries between employers and workers; and loss of human agency in workplace decision-making. Second, existing regulations do not adequately address these risks, even in jurisdictions with robust protections. Third, a range of policies can serve as interlocking elements of a regulatory strategy for addressing these risks, including: prohibitions on specific practices, including automated termination; restriction of legal bases; individual and collective notice obligations and data access rights; rights to explanation and human intervention; and impact assessment and information and consultation obligations.

We hope the presentation can support discussion of: whether these policy proposals are sufficient, especially in different cultural contexts; how they can be implemented in different jurisdictions; and what international action might help address the risks posed by algorithmic management.

The spread of digitization, automation and the emergence of platforms constitute an important source of changes in the organization of companies, employment relations as well as the labor market. The algorithms underlying these processes are increasingly determinants of social order, work organization and corporate decisions. The analysis of the impact of digitalization therefore implies the study of new economic relations that arise through 'datafication' and the use of algorithms and artificial intelligence (AI) via digital networks or are enabled or exacerbated by these technologies. AI-based mechanisms and the increasing reliance on algorithms to take decisions in both work organisation and HR are likely to involve an intensification, standardisation, and optimisation of the production process. These instruments provide new tools for companies to surveil and monitor employees and their performance. As decisions may be taken by “the algorithm”, transparency and accountability in managerial processes are reduced, undermining workplace participation and employee involvement. The combined effect of intensified control and intransparent decisions can lead, in the absence of adequate institutions and regulations, to an erosion of industrial democracy. The extension of AI and Algorithmic Management (AM) at the workplace level in all EU countries and its disruptive impact on employment relations contrasts with a diverse regulatory response. The diversity of forms in which AI and AM are applied across companies and sectors calls for flexible approaches, where collective bargaining must play an important role. However, governing and regulating AI through collective bargaining not only requires social partners to share a common diagnosis of the problems and solutions to them, but also having the skills and capabilities to implement them.

The challenges to employment relations posed by the extension of AI and AM by companies have led to growing demands from unions to regulate its use. This research examines the role of collective bargaining and employee participation mechanisms in regulating the use of artificial intelligence and algorithms at the workplace level and rendering it more inclusive and transparent. This is done through a comparative analysis of institutional developments at EU-level as well as in four countries belonging to different industrial relations models (Denmark, Germany, Hungary, and Spain). The study maps regulatory developments and gaps in response to these challenges, paying attention to the role of collective bargaining in governing the use of Artificial Intelligence and algorithms in the workplace.

Adoption of new Artificial Intelligence (AI)-based tools in workplaces is rising. Launched at the end of November 2022, ChatGPT, an AI-based tool which offers a range of functions from generating human-like responses to coding, is bringing public attention to the expanding potential of AI to transform professional jobs. These new technologies may eliminate many tedious tasks and expand worker voice and power through broad access to new tools to complete complex tasks. However, AI can also widen inequalities by deskilling or eliminating existing jobs and by concentrating control, knowledge, and skills among a small number of users.

This paper aims to analyze how ownership models and institutions supporting worker voice influence management approaches to reorganizing work and managing workers using AI-based tools. We focus on the game development industry, as this sector has been pioneering innovative approaches to apply AI-based tools in the workplace, and has been a target of worker activism aimed at improving worker voice due to often unstable hours and precarious conditions. Software engineers, programmers, and testers are experiencing significant changes to their job content and skills, as AI and algorithms are used to automate coding, software testing and maintenance, and to monitor remote employees. Our research design is based on a matched pair case study comparison across the US and UK. In each country, we will conduct semi-structured interviews with managers, employees, or worker-owners, and (where relevant) union representatives at one traditional non-union company, one worker-owned cooperative, and one unionized company. Archival data, company reports, and expert interviews will support findings.

Our analysis will examine differences in work organization, skills and training, and performance management practices across the cases, with a focus on the changes in each area connected with AI- and algorithm-based tools. Our central concern is to evaluate the impact of different forms of collective worker voice on each area; and their repercussions for job quality - particularly worker control over their job content, tasks, and working time. We expect to find more cross-national similarities in the traditional company cases, while outcomes at the union and worker-owned cases should be affected by industrial relations institutions and local ‘ecosystems’ supporting worker-owned cooperatives.

Our findings will contribute to the literature on worker voice, the impacts of AI on workplaces, and the future of work, and how institutional and policy measures can foster decent work in the increasingly international ICT and game development industries.