Bureaucratic Perspectives on AI in Public Service Delivery: An Experimental Approach

• In Event: AI Adoption in Public Management and Leadership

Abstract

Artificial intelligence (AI) has been integrated into both public and private organizations, resulting in transformations in how information is stored, analyzed, and shared among stakeholders. These transformations have presented public administrators with new promises and perils that they must navigate to determine how to extract the benefits of AI while mitigating the associated risks. Amidst this, scholars articulate a growing concern regarding the necessity for reforming organizational structures to accommodate the transformative decision-making capabilities enabled by AI. One of the most noteworthy changes to organizational structure that AI is likely to influence pertains to the traditional role of the bureaucrat. In particular, as AI use becomes more prevalent in public service delivery, nonexecutive bureaucrats will require training to effectively integrate both new AI tools, especially amidst the evolving concept of AI agents and the ‘artificial bureaucrat’.

While existing scholarship has examined factors that drive users' adoption of AI in public service delivery, little is understood about the bureaucrats’ motivations. We argue that bureaucrats play a critical role in shaping the adoption of AI in public service delivery. Decisions on AI implementation, scope, funding, and regulation in these services often originate from their desks. We therefore explore the factors that explain bureaucrats' support for AI in public service delivery. What characteristics of bureaucrats shape their assessment of public service delivery that includes AI? Further, how do different applications of AI as a policy tool influence bureaucrats' support for its use in public service delivery?

We argue that bureaucrats (compared to non-bureaucrats) are more likely to support the use of AI in public service delivery. Further, current research on bureaucratic decision-making highlights public service motivation (PSM) as a factor that shapes bureaucrats’ decision to support technology in public service delivery. This is also supported by a broader body of work on civic technology, which purports that technology adoption is partially amplified when it advances civic outcomes. We leverage these insights to argue that the use of AI in public service delivery is one instance where this motivation may be realized. Given the complexities associated with AI as a relatively new technology, we argue that bureaucrats' perceived administrative efficacy shapes the extent to which they support AI use in public service delivery.

Finally, AI applications are diverse, and individuals’ attitudes toward these uses are not linear or static. We argue that the specific ways AI is utilized in public service delivery play a significant role in shaping bureaucrats’ support for its adoption in this domain. Taken together, we argue that bureaucrats matter, and the mechanisms that elucidate how they exert their influence include (1) their assessment of their individual characteristics, such as public service motivation and perceived administrative efficacy, and (2) their evaluation of the policy use of AI.

     We use two vignette experiments, grounded in real government programs, to probe these assumptions. One experiment focuses on public service delivery in a welfare context, specifically examining an Emergency Rental Assistance Program, while the second centers on public service delivery in tax filing procedures.

Authors

• Donavon Johnson, Florida International University
  Non-Presenting Co-Author

• Lerone E Laing, Florida International University
  Presenting Author