As AI reshapes public service, frontline bureaucrats find their roles shifting. Hertie School PhD researcher, Anna Kubaszewska, explores how AI impacts the role of street-level bureaucrats, their freedom of judgement, and their routines. This post proposes avenues for future research to understand the dynamics between human discretion and AI in public sector decision-making.
Reliance on advanced technologies such as Artificial Intelligence (AI) to implement policies is changing the role of street-level bureaucrats at government agencies. Can human judgement actually be programmed? What will happen to the daily routines of these frontline workers who act as a link between the government and the public? Through their freedom of judgement (discretion), they are expected to balance detailed regulations with the complex reality of daily life. To manage their work, they create routines to deliver services, even when standards for service quality, volume, and specific goals are not clearly defined.
The rise of advanced technologies is shifting the discretionary activities of traditional street-level bureaucrats onto system-level bureaucrats, also referred to as data professionals. These professionals work with algorithms and data, and their roles entail multiple responsibilities in the data science realm. As opposed to bureaucrats who must integrate and protect public values when dealing with individual cases in real life, data professionals must safeguard these values when developing or working with algorithmic systems. They are expected to make conscious, value-guided decisions to overcome the often biased algorithms.
However, studies also point to the possibility of a more multifaceted scenario, where technology does not entirely eliminate street-level discretion. In many cases, street-level and system-level bureaucrats coexist with flexible boundaries between them and automation aids or replaces only part of the decision-making process. Depending on the complexity of the tasks performed by bureaucrats, this human role can take different forms.
Low-complexity decisions, like driver’s licence renewal or parking ticket issuance, will likely involve a limited level of human participation, which would serve as a last resort in emergencies. In the case of moderate to high-level complexity, including e.g., child welfare case management or grant funding decisions, technology is likely to be used as a tool that supports and assists public-facing bureaucrats. Therefore, they still retain some flexibility in their decision-making although this autonomy is neither apparent nor openly addressed.
Can human judgement actually be programmed?
Scholars are currently debating whether emerging technologies enhance or constrain the discretionary power of these bureaucrats. On the one hand, AI and algorithms allow quick access to predictive insights or risk profiles and can enhance human judgement. On the other, they can limit bureaucrats' involvement by increasingly automating decision-making. Additionally, a less explored but emerging perspective suggests a third outcome: street-level bureaucrats may try to resist the impact of technology on their work and ultimately on their discretion. Due to possible flaws and the difficulty of tailoring such systems to an organisational structure, as well as the complexity of social issues, bureaucrats are finding different coping mechanisms to deal with the challenges these tools have introduced. This has created a power dynamic incentivising people over technology.
Orlikowski (1992), argues that technology users can theoretically always control their tech interaction, no matter how inflexible it is or difficult to adapt to. And they can always opt out of engaging with it. She describes the role of technology within an organisation as a dynamic interaction between human agents, the technology itself, and the organisation's structures. These elements influence each other, shaping how technology is understood and used within the organisation.
She bases her assumptions on past evidence of emerging technologies like power plants, where some workers either refused to adopt them or obstructed their operation. These actions shaped how the technologies were used and often led to different outcomes such as underutilisation or tensions between parties involved in their adoption. Therefore, as it stands for now, the extent to which human judgement is being programmed is bound to the organisational context, the tasks for which it is used and the willingness of bureaucrats to engage with it.
What happens to the routines of street-level bureaucrats?
Despite existing research on how AI affects bureaucrats' roles and their freedom of judgement, relatively little is known about how it changes their routines. There is a need to understand how AI, through influencing the roles and discretionary power of street-level bureaucrats, alters their behaviour and how this, in turn, influences the informal rules and policies they create. Even though AI technologies are helping to ease the time and money constraints, these are being substituted by constraints related to the complexity of the new tools and the personal resources of bureaucrats.
One potential area of research is to explore the impact of AI on the routines of street-level bureaucrats and consider the influence it has on their roles and freedom of judgement. This would help assess whether the influence of bureaucrats within public sector organisations outweighs the built-in capabilities of technology in decision-making and operational processes. Such research could also explore how the autonomy of bureaucrats translates to their behaviour, and their conceptions of work and clients when AI becomes part of their daily routine.
This research could take the form of a case study in a place where the technology just has been introduced, to see how the routines of bureaucrats change initially. The other approach is to conduct a comparative study of a large number of organisations already using AI, to better understand how AI has impacted bureaucrats' behaviour and their conceptions of work and clients. Both of the approaches would shed light on the power dynamics of incentivising people over technology, as well as how street-level bureaucrats try to resist the impact of technology on their decision-making discretion.
Photo by Wesley Tingey on Unsplash