Article: Thursday 19 February
When artificial intelligence is introduced at work, the biggest decisions are often already made. By the time workers feel the effects on schedules or job content, the technology can be hard to reverse. The Dockers’ AI Toolkit, developed for the International Transport Workers’ Federation (ITF) by José Luis Gallegos, PhD candidate at Rotterdam School of Management and visiting fellow at Cornell University, is designed to intervene much earlier.
The toolkit offers practical guidance for employees, labour advocates, and particularly trade union members, on how to negotiate the use of AI before it reshapes work. Although rooted in the port sector, its relevance extends beyond dock work: AI is increasingly used to allocate tasks, monitor workers and support organisational decisions. “AI at work should not be something workers are forced to adapt to after the fact,” Gallegos says. “It should be something they help govern from the outset.”
Ports have long been testing grounds for new technologies, from containerisation to automated cranes and digital planning systems. In his PhD research, Gallegos examined earlier waves of automation at the Port of Rotterdam and observed how unions played a decisive role in ensuring that the introduction of automation did not result in increased inequality or job displacement. AI, he argues, raises the stakes further because it also reshapes decision-making. It influences planning, performance evaluation, and workflow coordination, affecting office-based roles as well as work on the terminal floor.
A key strength of the toolkit is that it looks at AI as a whole, rather than as a series of separate problems. Instead of dealing with job loss, surveillance, bias, and data use one by one, it brings them together in a single framework. That matters because the same AI system can be used in very different ways. It can be set up to improve safety, cut costs, increase efficiency, or tighten control over workers.
These decisions are typically made early—when systems are procured or designed—yet worker representatives are often not involved at that stage. The toolkit therefore concentrates on intervening while objectives, boundaries, and accountability structures are still open for discussion. The toolkit breaks negotiations into clear steps and includes model clauses that unions can adapt for collective agreements. “Negotiating AI with worker representatives is not about blocking or slowing technological change,” Gallegos emphasises. “It is about collectively setting the rules under which it operates.”
The guide is clear that simply informing workers after decisions have been made is not enough. When AI changes how work is organised, late consultation offers little real protection. Instead, the toolkit argues for firm rights that give worker representatives a real say both before and after AI is introduced. This includes the right to approve or reject high-risk systems and the ability to bring in independent experts who can assess technical, legal or ethical risks. The toolkit also calls for clear limits. Certain uses of AI should not be allowed at all, such as biometric surveillance, emotion-recognition tools or fully automated decisions that affect people’s jobs.
AI systems do not stay the same once they are introduced. They learn from new data, are updated by suppliers and are often used in more ways than first intended. To deal with this, the toolkit proposes setting up joint technology review committees, with equal representation from management and workers. These committees would regularly review the effects of AI systems, keep an eye on how they are used in practice and arrange checks when problems arise.
Another key contribution of the toolkit concerns worker data. Every time workers adjust a system or correct an error, they generate data that can be used to train AI. Over time, this data reflects their experience and practical know-how. As the toolkit notes, workers are “hired to move boxes, not to train AI”. Yet their knowledge can end up feeding systems that later deskill or replace parts of their work. For this reason, the guide treats data as a form of labour and explores ways workers could be compensated, for example through wage premiums, agreements on how data may be used, or jointly managed transition funds.
Although developed for dockworkers’ unions, the toolkit offers broader lessons for organisations introducing AI at work. This is especially relevant as new regulatory frameworks in Europe, including the EU AI Act, and similar initiatives elsewhere begin to clarify responsibilities for organisations deploying AI in employment settings. In this context, AI implementation is increasingly understood not only as a technical decision, but also as one that carries organisational and governance implications.
You can find the toolkit here.
Official citation: Gallegos, J. L. (2025). Dockers’ AI Toolkit: Future of Work
Science Communication and Media Officer
Rotterdam School of Management, Erasmus University (RSM) is one of Europe’s top-ranked business schools. RSM provides ground-breaking research and education furthering excellence in all aspects of management and is based in the international port city of Rotterdam – a vital nexus of business, logistics and trade. RSM’s primary focus is on developing business leaders with international careers who can become a force for positive change by carrying their innovative mindset into a sustainable future. Our first-class range of bachelor, master, MBA, PhD and executive programmes encourage them to become to become critical, creative, caring and collaborative thinkers and doers.
