Public Administration Paradigms in the AI Era: Revisiting and Reimagining

The views expressed are those of the author and do not necessarily reflect the views of ASPA as an organization.

By Tong Chen
January 23, 2026

From process automation to algorithmic welfare allocation, AI technologies are gradually integrated into public service delivery and policymaking. Yet classic public administration paradigms, designed for analog bureaucracies, are not sufficient to address new challenges such as algorithmic bias, privatization of governance and accountability gaps. A critical question arises: does the AI era demand a fundamental reimagining of a new paradigm?

Revisiting PA Paradigms in the AI Era

1. Politics/Administration Dichotomy: The Myth of AI Neutrality

Wilson’s model separated policy making (political) from implementation (administrative), assuming bureaucrats could act as neutral executors. Yet algorithms may appear objective, but they can still encode political and normative judgments into their design. These judgments are shaped by choices related to training data, manual weighting of variables and risk threshold calibration. For instance, predictive policing tools like PredPol, trained on historically biased arrest records, reproduce racial disparities under the guise of technical neutrality.

2. Scientific Management: Efficiency at What Cost?

Taylor’s scientific management sought to optimize workflows through standardization and hierarchy, a vision seemingly realized by AI’s ability to automate tasks. However, AI’s “efficiency” often sacrifices empathy and equity. Consider Amazon’s scrapped hiring algorithm, which penalized resumes mentioning “women’s” organizations, perpetuating gender disparities. While AI can empower frontline workers with real time data, it risks reducing public service to transactional interactions, stripping away the human judgment essential for equitable governance.

3. New Public Management: Accountability Erosion

New Public Management’s market centric logic exacerbates AI’s accountability gaps. Outsourcing algorithmic governance to private vendors creates accountability vacuums where neither governments nor contractors fully own the harms caused by AI systems. For example, Michigan’s AI welfare fraud detector falsely accused 40,000 residents (2013–15). Although the state ultimately paid $20 million in settlements, the case against vendor SAS Institute was dismissed for lack of subject matter jurisdiction. Similarly, Clearview AI’s facial recognition led to Robert Williams’ wrongful arrest in 2020, yet no penalties were imposed on the firm.

4. New Public Governance: Collaboration in the Shadow of Tech Giants

New Public Governance emphasizes collaboration across sectors, but AI infrastructure is increasingly monopolized by tech oligopolies. Toronto’s failed Quayside smart city project, led by Alphabet’s Sidewalk Labs, exposed public distrust in privatized data governance. New Public Governance’s ideal of inclusive collaboration falters when AI tools are gatekept by corporations or exclude marginalized groups digitally. Additionally, algorithmic systems can inadvertently marginalize groups lacking digital access or literacy, undermining this paradigm’s commitment to inclusivity.

5. Digital Era Governance: Citizen Centricity Meets Algorithmic Opacity

Unlike foundational digital technologies such as databases and online portals that streamline processes, AI introduces autonomous systems whose logic is often inscrutable. For instance, the United Kingdom’s 2020 A level grading algorithm and Australia’s Robodebt scandal deployed opaque AI tools that produced unexplainable and discriminatory outcomes. The “black box” undermines Digital Era Governance’s citizen centric transparency goals, as citizens cannot scrutinize decisions affecting their lives.

Reimagining a New PA Paradigm

1. Hybrid Human AI Collaboration

AI should augment, not replace, human judgment. Administrators need training to interpret AI outputs critically. Frontline workers must retain discretion to override algorithmic recommendations, ensuring context sensitive decisions. For example, South Korea’s AI Basic Act requires high impact AI systems in healthcare, employment and essential services to integrate human oversight mechanisms, user notifications and risk management protocols, ensuring automation aligns with human agency in consequential decisions.

2. Enforcing Accountability Through Legal and Institutional Mechanisms

Binding liability regimes must hold developers, vendors and users jointly responsible for AI related harm. Independent oversight bodies should audit compliance and investigate complaints. New York City’s Local Law 144 exemplifies this approach by mandating annual bias audits of automated hiring tools, requiring public disclosure and imposing penalties for noncompliance. Public agencies should also embed accountability in procurement contracts by requiring vendors to provide transparency reports, comply with fairness audits and accept liability for AI driven errors. Without legal teeth, ethical AI frameworks remain hollow.

3. Democratizing AI Infrastructure

To counter tech oligopolies, governments must invest in public AI infrastructure such as open source tools, municipal data cooperatives and digital literacy programs. Estonia’s X Road system, which integrates public services under government controlled data exchange, offers a model. Collaborative platforms should co design AI with communities, ensuring marginalized voices shape algorithms affecting their lives.

4. Institutionalizing Algorithmic Stewardship

Algorithmic stewardship refers to the proactive oversight of algorithmic systems. This requires continuous monitoring and sunset clauses to remove obsolete systems, as reflected in the European Union’s AI Act, which enforces risk tiered rules and bans harmful applications. Algorithmic stewardship also calls for cultivating technical and ethical expertise. Governments must train civil servants in data analysis, AI oversight and ethical discernment. South Korea’s AI National Strategy, which includes nationwide upskilling for public officials, offers a strong example.

Conclusion

The AI era demands a paradigm shift that transcends incremental adaptations of past models. The new paradigm must offer a holistic framework to harness AI’s potential while safeguarding accountability, transparency and fairness. The urgency of this transition cannot be overstated. Without a coherent paradigm, AI risks entrenching technocratic control, exacerbating inequities and eroding public trust.

---

Author: Tong Chen is a Ph.D. candidate at Rutgers School of Public Affairs and Administration (SPAA). Drawing on computational methods and large-scale data analysis, his research focuses on AI & disruptive technology governance, public corruption, and social media communication.

(2 votes, average: 5.00 out of 5)
Loading...