5 Key Thoughts on Platform Plugins & AI Regulation

1. Black box or one-way mirror?

Algorithms are used to determine access to work, working conditions, the nature of work and its evaluation. They are often described as a ‘black box’: a closed system in which decisions are made automatically, based on variables selected by the technology company that developed it. But is ‘black box’ the right term?

The metaphor suggests that the algorithm is inherently difficult to open and understand, but this is just one single perspective. Black box describes the perspective of those who cannot see the choices and data behind an algorithm. For the people who create and manage it, there should be no mystery. Not knowing how an algorithm works is a serious business risk, as I argued during a workshop on the ‘algorithm accountant’ in 2019. The one who controls the algorithm is in control of the data used and the variables and parameters of the algorithm.

Perhaps a better metaphor is a ‘one-way mirror’. Think of an interrogation room in a movie: the suspect cannot see the detectives watching from the other side, but they can see everything the suspect does. This captures the power imbalance at the heart of many algorithmic systems: one party remains invisible while closely observing the other.

What if we stopped talking about ‘black boxes’ and started talking about ‘one-way mirrors’ instead? The debate would shift from the technical challenge of opening an obscure box to the real issue: unequal access to information and the power to hold decision-makers accountable. The administrator would not need to open the box. They would simply need to take you into the control room.

2. Complexity as a defence mechanism

Complexity is often used to justify why something cannot be done. I was reminded of this at a platform co-operative conference in Rio de Janeiro in 2022, where I presented GigCV, a service I developed that allows work platforms to share workers’ data through a digital CV. At least half the audience were lawyers, and someone asked how I dealt with the legal complexity. My answer was simple: “GigCV is legally straightforward. Complexity is often used as an argument when you don’t want to do something.” The audience’s reaction suggested this struck a chord.

Complexity is also often used to secure an advantage for one side. One example of this is the pricing mechanisms used by taxi platforms. This has evolved from a simple combination of distance and time to complex systems where drivers and customers can be offered different prices for the same journey. Who benefits from this complexity, and who bears its costs? We need to question whether its benefits outweigh its drawbacks and examine who gains and who loses. Bringing different stakeholders and disciplines into the debate can help.

3. The trade union as a beacon of hope

Trade unions are often expected to protect workers as management of work becomes increasingly automated. This makes sense: as work becomes more fragmented and power and information imbalances grow, collective organisation becomes even more important.

I have often criticised Dutch trade unions for their role in this debate. But recent conversations raised another question: are we expecting too much from them? Negotiating the impact of technology requires specialist expertise. My research into platform cooperatives revealed a similar challenge: we cannot expect a skilled taxi driver to suddenly become a successful tech entrepreneur. Why, then, should we expect trade unions to become technology experts overnight?

Their challenges are structural, too. Many workers affected by the gig economy are self-employed, while unions in the Global North have traditionally focused on employees. The sectors concerned have also often been fragmented and less visible to unions.

Trade unions need greater technical expertise to negotiate effectively with technology companies, develop targeted solutions and potentially build tools that strengthen workers’ digital resilience. They can learn from grassroots initiatives and collaborate with data-driven organisations such as Worker Info Exchange. Governments and legal experts across disciplines also have a role to play.

Ultimately, we need to stop treating trade unions as solely responsible for solving these problems. Addressing the impact of technology on work requires collaboration, expertise and shared responsibility.

4. Platform plugins

One possible solution for implementing regulations more effectively and giving representatives more opportunities is to consider how the platform’s technology can be used to promote compliance and co-determination. In my work at the Amsterdam University of Applied Sciences I explore ‘platform plugins’ as tools for implementing regulation and sharing data between platforms, governments and other stakeholders.

One project is GigCV, which is a central data portability tool that connects platforms to workers through APIs, giving them access to their data in a digital CV. Currently used by three platforms, it has provided over 100,000 workers with access to their data and generated more than 34,000 API calls. Its central infrastructure makes it easy for platforms to share data with their workers and scaling relatively straightforward while keeping governance centralised.

This connects to another concept I have planned to develop: ‘Wage Against the Machine’. The idea is to use a prompt generator informed by living-wage data to ensure that when algorithms or AI agents hire workers, they do so without exploiting them. It builds on ideas discussed with the Swedish trade union Unionen, which recognised years ago the potential of encoding collective agreement terms directly into platform algorithms.

5. Platforms for access to workers

Platforms centralise fragmented markets. As a ‘hub’ they can be used as an access point to reach and influence workers that were previously invisible and unreachable. A project I am working on is a design solution for the implementation of Article 20 of the European Platform Work Directive.This legislation requires platforms to provide communication channels for workers and their representatives. Rather than requiring thousands of platforms to build and maintain separate channels, I propose a shared public infrastructure to which platforms connect via APIs, supported by multi-stakeholder governance.

To conclude

In a world where the impact of technology on work is set to increase and workers’ information position is set to diminish, it is important to take a critical view of the language used in the debate and to be more aware that the speaker represents only one perspective. To improve the position of workers, traditional solutions will simply no longer always work. This presents an important challenge for workers’ representatives and policymakers to act. Because technology also offers many opportunities to strengthen the position of workers in ways that were previously unimaginable.