Unclear regulations are challenging the use of artificial intelligence in the public sector. Two legal scholars propose that new legislation providing a clearer legal basis for the use of AI in public administration should be enacted to provide clarity and strengthen both innovation and citizens' legal rights.
The government aims to make Denmark a world leader in the application of artificial intelligence. Specifically, politicians have set a target of saving 30,000 full-time equivalents by implementing the technology. However, many of the AI projects in which local authorities and government bodies have invested millions never become operational.
According to a new study by researchers at the Faculty of Law, this is not necessarily due to the technology itself, but rather to the lack of a clear legal framework. Denmark therefore risks hindering its own ambition to become one of the world's leading countries in the use of artificial intelligence in the public sector.
'There is a strong desire to adopt AI in the public sector, but also uncertainty regarding the legal framework for its use. From a legal certainty perspective, it is therefore important that there is clarity and predictability regarding when and how AI can be used, so that both public authorities and citizens can have confidence in the way the technology is utilised,' says Aysel Küçüksu, assistant professor at the Faculty of Law, University of Copenhagen, and a fellow at CAISA - the National Centre for AI in Society.
Drawing on the concept of the 'digital graveyard', the researchers describe how promising AI projects never make the transition from development to practical application.
Seven out of ten projects never went live
The study is based on legal analyses of Danish and European legislation, the practices of the Danish Data Protection Agency, and experiences from a number of Danish flagship AI projects in the public sector. The researchers point out that only six out of 22 completed flagship projects were put into operation, whilst the rest were closed without implementation. Legal uncertainty regarding the legal basis was one of the main reasons.
- Denmark lacks a comprehensive legal framework for the use of AI in public administration when processing personal data.
- Many public AI projects have been halted due to uncertainty regarding the legal basis.
- Only six out of 22 completed Danish flagship AI projects have been put into operation.
- Norway and Finland have already updated their legislation in this area.
- The researchers recommend a Danish framework law for artificial intelligence in the public sector.
The problem arises particularly when public authorities wish to use AI systems for case processing involving personal data. In this context, many public authorities find it unclear exactly what legal basis is required.
'Denmark currently has a fragmented regulatory framework, where individual sectors have specific rules, whilst others lack a clear legal basis. This creates uncertainty and makes it difficult to work systematically with artificial intelligence,' says Henrik Palmer Olsen, professor at the Faculty of Law, University of Copenhagen.
Inspiration from Norway and Finland
The researchers have compared the Danish situation with developments in Norway and Finland, both of which have updated their administrative legislation to establish clearer rules for automation and artificial intelligence in the public sector.
According to the researchers, experience shows that it is possible to create a more coherent framework without compromising citizens' rights. They therefore recommend that Denmark introduce a comprehensive framework law for artificial intelligence in the public sector. Such a law must clarify when AI may be used, what requirements apply regarding transparency and proportionality, and how citizens' personal data is protected.
'There is a need for a set of policy rules that regulate the use of the technology whilst protecting citizens' fundamental rights by ensuring proportionality and transparency,' says Aysel Küçüksu.
AI is not the solution to everything
The researchers also emphasise that artificial intelligence is not always the right solution.
Some projects have been abandoned because the data set was insufficient, because the error rate was too high, or because the technology was not yet mature enough for the task. Authorities should therefore not only ask whether AI can be used, but also whether it should be used.
'Responsible digitalisation is not about introducing artificial intelligence in as many places as possible. It is about identifying the tasks where the technology actually creates value, and where legal certainty, transparency and professional quality can still be guaranteed,' emphasises Henrik Palmer Olsen.
The researchers hope that the findings will contribute to the current political debate on how artificial intelligence should be implemented in the Danish public sector in the coming years.
The study, published in the academic journal Politica, is entitled 'AI in the public sector: from Nordic experiences to a Danish law on artificial intelligence'. Read it here.
For their research and analysis, the researchers have used offentlig-ai.dk, Denmark's first database on the use of artificial intelligence in the public sector. It was developed on the initiative of Assistant Professor Aysel Küçüksu and Senior Lecturer Zuzanna Godzimirska and is hosted by the Faculty of Law.
