What happens when artificial intelligence is used in public administration?

In recent weeks and months, artificial intelligence has risen high on the Norwegian public agenda.

What is happening with ChatGPT in school? Who is responsible for regulating this technology? Can machines really think?

But even though artificial intelligence is now a favourite topic for columnists around the country, this interest in AI did not arise overnight.

In connection with my doctoral work, I have spent four years looking at how and why the Norwegian public sector has adopted, or envisages adopting, technologies such as artificial intelligence, and how this fits into the larger context of technological and societal change.

• You might also like: Data is never objective. Sometimes we must give free rein to experience and intuition

Beyond previous data usage

Data has always been a central part of the modern state. Good decision-making and the construction of a welfare state depend on detailed and up-to-date information about its citizens.

But something is about to happen that goes far beyond the use of data for statistical, case management and archiving purposes.

Making the public sector data-driven will rely on more, different or new data, and will make use of data using technologies such as data platforms or machine learning.

Central to this change in the administrative apparatus is the idea that data should become a central resource for all forms of knowledge production in the public sector. This is not always about automation, big data or AI, but an interaction of new data practices in the public sector.

Will make the public sector more efficient

Artificial intelligence is thus a part of a major change in government administration that aims to make the public sector more efficient with the help of data; or as was stated in a 2019 digitalization conference: “The country will be built with data”.

It can often seem as if AI is the solution to all challenges facing the public sector, but this assumption proves difficult to realize in reality. This is due to both the practical challenges of using and implementing the technology, and the consequences these changes have for the relationship between citizen and state.

A number of researchers in the new and interdisciplinary field of critical algorithm and data studies see data-driven public administration as a paradigm shift. But paradigm shifts bring with them a number of new questions and challenges.

When we give the state the opportunity to compile new, larger amounts or different data, this increases the  ability of the public sector to see, predict, control and manage citizens’ behaviour. It shifts the balance of power between citizen and state.

This is not necessarily a bad development, but deserves a critical look from more than the technologists and lawyers.

Changes the relationship between citizens and the state

Because how does the use of more, different and new data change the relationship between citizen and state? Who is allowed to decide what a data-driven administration should look like? What assumptions is this paradigm shift based on? What role do fundamental values in the Norwegian welfare model play in technological change?

These are big, challenging questions that in most cases require more than a PowerPoint or a newspaper article to answer.

The public sector is different from the private sector

In the research for my dissertation, I found, for example, that the Norwegian public sector often encounters a number of challenges when AI is to be developed or implemented in practice.

Technology development in the public sector differs fundamentally from technology development in companies such as Google or Twitter, in that the public sector is set up to prevent major shifts in power between citizens and the state and to safeguard basic rights and values in the Norwegian welfare model.

This can be experienced as very frustrating for technologists, but perhaps these challenges can also tell us something more about how the data-driven paradigm shift changes the foundations of the welfare state.

• You might also like: The fear of new technology in schools is exaggerated

Most of questions remain unanswered

Most questions about how the challenges  AI poses to the public sector should be addressed remain unanswered for the time being. But many researchers point to the fact that greater public involvement and participation is absolutely crucial here.

Although strategies and parliamentary messages promise to put the “user at the centre,” the citizens’ voice is often absent when these documents are developed, or in the public debate around the topic.

Technological development is seen as something apolitical and a “natural development” in the Norwegian context.

But many see the large amounts of data kept in records and data warehouses  as a gold mine.

At the same time, scandals across Europe, linked to AI in public administration, have already shown that the idea of data-driven public administration is also about power, democracy. and social change.

Reference: Reutter, Lisa Marie. Datafication of Public Administration: Between Policy and Practice. Doctoral thesis. NTNU Open.