đŹ Research summary by Angshuman Kaushik, Researcher in AI Policy, Governance and Ethics.
[Original paper by Anneke Zuiderwijk, Yu-Chen Chen and Fadi Salem]
Overview: The expanding use of Artificial Intelligence (AI) in public governance worldwide, has not only opened up new opportunities, but has also created challenges. This paper makes a systematic review of existing literature on the implications of the use of AI in public governance, and thereafter, develops a research agenda.
Introduction
There is no denying the fact that AI has been used for quite some time now, and its use has resulted in both positive and negative outcomes. Further, considering its scope, AI is a multidisciplinary area of research, rich with a vast number of papers pertaining to its myriad applications. Within that extensive gamut, the emphasis of this paper is on the literature that addresses the effects of the uses of AI in the public governance setting. This paper narrows down its focus on the articles that research on the implications of AI in the context of public administration, digital government, management, information science and public affairs. It deals with the issues relating to fairness, bias and governance questions pertaining to transparency, and regulatory frameworks. For instance, how does the implementation of specific AI technologies affect accountability of government institutions? As far as the research approach is concerned, the researchers conducted a systematic literature review of the relevant material. After an extensive review, the researchers found a number of potential benefits and challenges relating to the use of AI in public governance enumerated in them.
They identified the benefits in nine categories: 1) efficiency and performance benefits, 2) risk identification and monitoring benefits, 3) economic benefits, 4) data and information processing benefits, 5) service benefits, 6) benefits for society at large, 7) decision-making benefits, 8) engagement and interaction benefits, and 9) sustainability benefits. In addition to the potential benefits, they also identified eight challenges of the use of AI in their literature review, which are divided into eight categories: 1) data challenges 2) organizational and managerial challenges 3) skills challenges 4) interpretation challenges 5) ethical and legitimacy challenges 6) political, legal and policy challenges 7) social and societal challenges, and 8) economic challenges.
Use of AI in public governance
An Example
One example relating to the application of AI in the governance-setting is the use of SyRI (âSystem Risk Indicationâ) by the Dutch Government to detect possible social welfare fraud. It had not only issues with transparency and a host of other factors, but the algorithm also turned out to be a âblack boxâ. Its operation was eventually brought to an end by the court for violating Article 8 of the European Convention on Human Rights (ECHR), which protects the right to respect for private and family life. The requirement of Article 8 is that, any legislation should strike a âfair balanceâ between social interests and violation of the private life of the individuals. (The intention of citing this particular example is not to portray the deleterious effect of AI, but to show the application of AI in governance, in general.) There are numerous such other cases, with beneficial outcomes pertaining to the use of AI in various sectors of the government.
Potential benefits
The use of AI in governance has massive implications for society, in general and individuals, in particular. The reason being, the administration, and its various functionalities have to directly deal with the masses, within their respective spheres of jurisdiction. Through this paper, the researchers have uncovered their findings with respect to a comprehensive review of 26 articles pertaining to the use of AI in public governance, which were published in the last 3 years. After analyzing the content of the articles, the researchers found that they contained a number of potential benefits of the use of AI in public governance. It was found that efficiency is improved by automating processes and tasks or by simplifying processes using machine learning. Further, AI aids in increasing monitoring of urban areas, fraud detection, law enforcement and enhancing the âsmartnessâ of the cities. The researchers also noticed that AI for public governance leads to economic benefits, such as making e-government services and systems more economical. Moreover, data and information processing benefits also accrue due to processing of large amounts of data in limited time. Another area where AI has a potential positive impact with respect to public governance is that it leads to improvement in the quality of public services. It also leads to the creation of public value, decision-making and sustainability benefits.
Challenges
Alongwith the potential benefits, the researchers also searched for the challenges of AI use in government, in the review of the 26 articles zeroed in on by them. They identified challenges relating to the availability and acquisition of data, organizational resistance to data sharing, limited in-house talent, complexity in interpreting AI results, ethical challenges, undermining the due process of law and effect on the labor market.
Going Forward â The Research Agenda
After an analysis of the various potential benefits and challenges, the researchers put forward a research agenda on the implications of the use of AI for public governance. It comprises eight process-related recommendations and seven content-related recommendations for researchers that examine the implications of AI use in public governance.
Process-related research recommendations
- Avoid applying AI-related terms superficially in public governance sources
- Move beyond the generic focus on AI in public governance sources
- Move to methodological diversity instead of dominant qualitative methods
- Expand conceptual and practice-driven research from the private to the public sector
- Increase empirical research on the implications of AI use for public governance
- Go beyond exploratory research and expand explanatory research
- Openly share the research data used for studies on the implications of the use of AI for public governance
- Learn from applicable pathways followed by digital government scholarship in its early phases
Content-related research recommendations
- Develop AI public governance scholarship from under-theorization into solid, multidisciplinary, theoretical foundations
- Investigate effective implementation plans and metrics for government strategies on AI use in the public sector
- Investigate best practices in managing the risks of AI use in the public sector
- Examine how governments can better engage with and communicate their AI strategic implementation plan to stakeholders
- Investigate a large diversity of possible governance modes for AI use in the public sector
- Research how the performance and impact of public sectorsâ AI solutions can be measured
- Examine the impact of scaling up AI usage in the public sector.
Between the lines
Although the paper has dealt with the subject very thoroughly, this is only the starting point of a âjourney of learningâ, which necessitates an iterative approach, involving relevant stakeholders. The findings matter, as it would enable the various actors involved in the process to have a better understanding of the issue in hand, and take appropriate steps, in the right direction, going forward. In my view, further deep dives into the application of AI in public governance at the grassroots level, through case studies, will yield specific insights. It deserves mention here that different cultures perceive AI and its outcomes in a different manner. Hence, more in-depth research, keeping in mind the cultural sensitivities, tastes and habits of different communities, would definitely, bring about a new flavor to the ever-growing field of AI, and its application, particularly, in the field of governance.