Outlook

Human-centred and trustworthy AI can make the lives of many easier – this applies to its use in the world of work, healthcare and social insurance. Nonetheless, the potential of AI appli­cations can only be realised if progress is made in terms of the availability and scope of data. A limited database is associated with the risk of bias and thus potentially incorrect decisions by AI applications, which is a risk that should not be assumed, particularly in the area of managing public services and benefits. There is therefore a broad consensus that responsibility within the scope of AI use must be traceable to human decisions and actions, which means that a human bears the ultimate responsibility, and AI is therefore merely a decision-making aid.1

AI should respect European values and protect people’s fundamental rights. To achieve this goal, data protection and non-discrimination must be guaranteed, transparency must be pursued, AI risks must be identified, and AI must be regulated accordingly. The AI Act is the starting point of such regulation. In line with the major task it is intended to fulfil, the AI Act is a very extensive project. By August 2026, more than 70 implementing and delegated acts will have to be drafted to lay down the general rules and provisions of the Act. The implementation deadlines are also very tight. Prohibited systems must be phased out as early as mid-February 2025. At the same time, the standardisation organisations are working on deepening trust and describing technical requirements with the help of norms and standards for AI. Standards should be used as a specifying instrument to recognise and reduce risks. With the AI Act coming into force, the work really gets started.

Closely monitoring the social impact is crucial when implementing the AI Act. Moreover, the social partners and civil society should be involved in the discussion about expectations and the use of AI in order to achieve the broadest possible participation and support. This also applies to the most efficient and effective organisation of governance structures. With this in mind, the DRV Bund has contributed to the development of a charter of values for human-friendly automation, according to which people must be at the centre of the use of AI.2

The European Commission is not standing still either. In her Mission Letter to Commissioner-designate Roxana Mînzatu, the European Commission President Ursula von der Leyen focused on the impact of digitalisation in the world of work and called on the Commissioner-designate to present an initiative on algorithmic management.3 Following initial sector-specific regulation with the Platform Work Directive, the planned initiative aims to improve transparency, fairness and accountability of algorithms in the world of work as a whole. The aim here will also be to use the new opportunities to optimise work processes without compromising the health and safety of employees.