AI and the EU Workplace

AI technology is poised to transform employment in a number of radical ways. This raises a set of significant challenges to the current system of labour law and data protection, from discrimination, to hiring and firing, to worker monitoring.

The European Union has introduced for negotiation an Artificial Intelligence Act which seeks to bring corporate use of AI in line with existing EU norms.

As widely discussed in the media, some experts have identified AI as the next ‘general purpose technology’, the fourth since the steam engine, electricity and semiconductors. As such, mapping today the full extent of its impact on economies generally and our working lives specifically maybe impossible.  However, a general outline of the most immediate applications can be glimpsed.

An example of one immediate concern is that these systems will increasingly set the pace of work and continuously monitor the work productivity of humans in their vicinity.

Overall, Industry 4.0 Automation systems reduce room for employees’ autonomy and increase forms of management control. Robots, wearable devices and other autonomous systems can collect or feed data to AI, enabling optimisation but also treating human workers as merely a biological sub-routine in an increasingly automated system, with machines setting the pace of work and large numbers of human beings being unable to compete in the new system.

The EU Approach and Objective

The EU overarching objective is for AI tools to be seen as complementing and enhancing human workers rather than displacing or controlling them. Businesses that seek to utilise AI (and they will have to, if they are to stay in the game) must do so in a fashion that incorporates workers existing fundamental rights.

Labour Market Impact

In labour markets generally, it is likely that AI tools are already, and will increasingly, have the largest and most immediate impact in the following areas:

  • Recruitment: The problem that AI has is that whilst it can sift through a massive quantity of data in very short order, it is nonetheless biased in its selection mechanisms. This could result in far more brutal selection mechanisms where the AI is selecting for only the top 2% of any applicant pool, purely on quantitative metrics. Where sifting is used prior to a human applicant advisor, the result is that whole categories of applicant are not even considered.
  • Performance Management: AI tools have the potential to transform data analytics on employee performance, enabling far more comprehensive attempts at fostering business efficiency. Overall, drawing on this literature, technology seems destined systematically to collect and present data about individuals and teams’ performances. However, the most recent reports from the field indicate that the capacity of these analytics to improve performance is quite limited, and human beings will still be required to address overall problems. People analytics can promote a false sense of certainty regarding the data, a tech solutionist fallacy.
  • Task Distribution, management and evaluation: One of the scenarios depicted by experts is for technology to be introduced with the aim of spurring worker productivity. Along the lines of what is happening in ecommerce warehouses, AI can be introduced in an attempt to boost production and productivity, as well as track ‘unionisation risk’. This raises the risk of what some are calling the ‘Amazonian Era’ emerging out of the platform or ‘zero-hours’ economy. This entails not the replacement of humans by machines but the treatment of humans as machines. At the same time, analytics of tasks could lead to a lightening of the workload, improving health and safety conditions, and higher generation of value through more effective deployment of workers.
  • Retention, rewards and promotion: Some companies already follow a ‘holistic’ digital management approach aimed at governing all areas of people’s work experience, ‘a panopticon’. These can culminate in a continuous feedback process, with the result that these processes can be used to determine workers compensation and in disciplinary procedures.

EU Legislation

The EU has concluded that AI-enabled recruitment and employee managerial tools should not be implemented lightly.  From a business perspective, this is because data accessibility in many workplaces remains poor, particularly when it comes to quality data. Furthermore, existing devices and work tools may poorly support the incorporation of AI features, potentially demanding significant additional investment, such has hiring large teams of AI experts. This will be beyond the resources of most but the larger businesses.

The EU believes this warrants regulation to ensure occupational safety, data protection, and other labour and employment rights are fully considered when designing these systems.

At present, three pieces of EU legislation will govern the interaction of AI with the existing economic system:

  • GDPR;
  • EU Directive 2002/14/EC (the general minimum requirement for employee rights to information and consultation); and
  • the EU AI Act – in EU Council negotiation.

In June 2023, the European Parliament approved the negotiation in the EU Council of the EU AI Act, which is the first specific attempt to regulate AI tools.  The aim is to reach agreement by the end of this year.

The Act looks to ban outright certain categories of AI usage like social scoring, cognitively manipulative AI software and real-time and remote biometric information collection (facial recognition).

It intends to require the following AI products to be registered on a database and be subject to periodic review:

  • Biometric identification and categorisation of natural persons
  • Management and operation of critical infrastructure
  • Education and vocational training
  • Employment, worker management and access to self-employment
  • Access to and enjoyment of essential private services and public services and benefits
  • Law enforcement
  • Migration, asylum and border control management
  • Assistance in legal interpretation and application of the law.


In general, the aim of all AI legislation should be to ensure that the AI systems work for the benefit of their human counterparts rather than creating a race to the bottom and an increasingly ruthless competition environment benefiting only a minority of human beings. It should also prevent the misuse of data and the codification of systematic bias into the labour market, as well as curb the ability of AI systems to engage in systematic deception.

Finally, it should seek to ensure that humans remain a part of the decision loop, and combat a tendency towards tech solutionism and algorithmic absolutism that misrepresents the capacities and shortcomings that these technologies have.