The paper is divided in two main parts. The first part introduces the agency theory and its application to two relevant aspects: the agency theory in the public sector and the agency theory involving artificial agents. The second part aims at providing answers to the research questions, by discussing the changes in the agency of the public administrations, as well as the changes in the control methods used to monitor these administrations. Finally, the conclusion summarizes the answer to the research questions, exposes the implications and limits of this paper and offers leads for possible future research on this topic.
Automated decision-making (ADM), a type of algorithm which supports decision-making and combines advanced analytics and data minig to make predictions, has been developed in various public sector fields, from predictive policing to healthcare, and is increasingly helping public agents by delivering analysis that they can leverage to make their decisions. This technique involves three main stakeholders: the programmer of the algorithmic system; the user, who is the public agent operating the ADM system; and the individuals affected by the decisions made using ADM. This paper focuses on the consequences on the governance and responsibility of administrations increasingly relying on algorithms to make their decisions. Does the introduction of ADM in public administrations transform their agency? If so, why does this change occur and how does it impact the control methods required to supervise the actions of administrations?
The chosen approach is the agency theory, which is suited to deal with delegation, specifically between actors from different contextual backgrounds. France has been chosen as the case-study for this topic, as it has put in place relevant laws and public institutions in order to deal with public ADM. The method chosen to investigate this issue is based on a literature review, as it is appropriate to approach a case-study. This includes scientific papers for the technical aspects, from computer sciences to social and political sciences, as well as reports from governments, international institutions and private companies. More general literature, such as articles and blog posts are used for information on the use of ADM in France and the public debate surrounding it. Finally, the methodology also includes semi-structured interviews led with experts working on the topic of ADM in the public sector.
Table of Contents
1. Introduction
2. State of research
3. Agency theory
3.1. Agency theory in political sciences
3.1.1. Theoretical context
3.1.2. Principal-agent problems
3.1.3. Control methods
3.2. Agency theory with artificial agents
3.2.1. Definitions and agency of algorithms
3.2.2. Principal-agent problem with artificial agents
3.2.3. Control methods
4. Analysis of Automated decision-making in the public sector
4.1. Case-study: France
4.1.1. Public institutions in France
4.1.2. Control methods
4.1.3. Algorithms in the French public sector
4.2. Changes in agency when ADM is used in the public sector
4.2.1. Potential risks of using artificial agents
4.2.2. Responsibility of decisions taken with the help of ADM
4.2.3. Consequences on the organisation of administrations
4.3. Changes in control methods when ADM is used in the public sector
4.3.1. Selection and development of the agent
4.3.2. Supervision and monitoring of the agent
5. Conclusion
Research Objectives and Core Themes
This master's thesis examines the impact of integrating automated decision-making (ADM) systems into public administrations, specifically within the French context. The primary research objective is to investigate how the introduction of artificial agents transforms the agency of public administrations and consequently affects the established control methods used to ensure accountability and alignment with public interest.
- Application of Agency Theory to artificial intelligence in the public sector.
- Analysis of institutional changes and shifting responsibilities in French public administration.
- Evaluation of control mechanisms, including transparency, explainability, and algorithmic literacy.
- Investigation of the "principal-agent" relationship between citizens, policymakers, and algorithmic systems.
- Critical assessment of the risks associated with automation bias and technical opacity.
Excerpt from the Book
4.2.1. Potential risks of using artificial agents
Although, for now, ADM systems in the public sector always involve the participation of a human and do not make decisions automatically, there is a risk of agency being delegated almost entirely to the algorithms. With too complex calculations and extremely large amounts of data being considered, it becomes impossible for civils servants to contradict the results of an algorithm. “A suggestion is never only a suggestion” and people tend to follow what the algorithm says, even if the decision is not meant to be taken automatically (Smuha in: LIBE Committee 2020, [03:08:15]). The expert in Interview 4 gives an example of this with the issue of town commissions attributing spots in nurseries. From 300 applications, an algorithmic system selects only 5 that the commission has to study in detail. Although this is not fully automated, it raises the question of where the real decision-power lies: in the algorithm or in the final commission? In such cases, the chain of individual responsibilities is very difficult to unroll (Interview 4 2020). The introduction of ADM also brings in an additional human actor into the decision-making process: the programmer. This can have consequences on the decision outcome, as programmers consciously or unconsciously insert in their systems their own moral choices and value patterns (Martini 2019, 48). Moreover, when algorithms are developed by private companies, public agencies risk losing control and sovereignty over their decisions (Castelluccia and Le Métayer 2019, 22). Finally, for ML algorithms, the Constitutional Council pointed out that, because they define their own rules, it cannot be guaranteed that they follow the law and there is the risk of administrations surrendering their regulatory power to such algorithms (Interview 1 2020, Interview 4 2020).
Summary of Chapters
1. Introduction: Introduces the rise of Automated Decision-Making (ADM) in public sectors, using predictive policing in Marseille as a primary example to highlight both efficiency potential and democratic concerns.
2. State of research: Outlines the academic landscape, bridging political science perspectives on bureaucracy with computer science studies on AI alignment and ethical algorithmic governance.
3. Agency theory: Establishes the theoretical framework by applying classical principal-agent dynamics to public administrations and subsequently adapting these models for interactions involving artificial agents.
4. Analysis of Automated decision-making in the public sector: Examines the French case study, analyzing how administrative structures, responsibility, and control methods are impacted by the adoption of algorithmic tools.
5. Conclusion: Summarizes how ADM transforms administrative agency and underscores the necessity for new transparency, accountability, and ethical frameworks to protect individual rights.
Keywords
Automated decision-making, ADM, Agency theory, Principal-agent problem, Public administration, Artificial intelligence, France, Algorithmic accountability, Transparency, Explainability, Machine learning, Digital transformation, Public sector ethics, Predictive policing, Governance.
Frequently Asked Questions
What is the core focus of this research?
The work investigates the introduction of automated decision-making systems in the public sector, specifically looking at how this technology affects agency, responsibility, and control mechanisms in French public administrations.
What are the primary themes discussed in the thesis?
The core themes include the application of agency theory, the shift in power dynamics between administrators and algorithmic systems, legal and ethical regulation of AI, and the challenges of accountability in an era of automated governance.
What is the main research question?
The research asks: Does the introduction of ADM in public administrations transform their agency? If so, why does this change occur, and how does it impact the control methods required to supervise the actions of administrations?
Which scientific methodology is employed?
The research utilizes a literature review covering technical, social, and political science papers, combined with semi-structured interviews with experts from institutions like Etalab, Capgemini, and the French National Council for Digital.
What is covered in the main section of the paper?
The main section provides a detailed analysis of the French context, examining how institutions utilize ADM, the potential risks of automation bias, and the evolving requirements for transparency and oversight.
Which keywords characterize the work?
The work is characterized by terms such as agency theory, automated decision-making, public administration, AI ethics, accountability, and transparency in government.
How does the introduction of ADM alter the principal-agent relationship?
The paper argues that ADM adds a new layer of complexity to the principal-agent relationship by introducing programmers as key influencers and creating situations where the agent's behavior (the algorithm) becomes difficult for the principal to understand or control.
What role does the programmer play according to the author?
The programmer is identified as an additional, crucial stakeholder who can consciously or unconsciously embed personal or corporate biases into the ADM system, thereby influencing the outcomes of public sector decisions.
Why is the French administrative system chosen as a case study?
France is chosen because it has actively implemented specific legislation and established specialized institutions, such as Etalab and the CNIL, to navigate and regulate the challenges posed by public sector algorithms.
What is the conclusion regarding administrative responsibility?
The author concludes that while legal responsibility remains with human decision-makers, ADM leads to a "dilution of responsibility," making it critical to establish clearer attribution models and enhance algorithmic literacy among civil servants.
- Quote paper
- Hortense Fricker (Author), 2020, Automated decision-making in the public sector. Artificial Intelligence vs Administrative Intelligence?, Munich, GRIN Verlag, https://www.grin.com/document/972247
