Governments increasingly classify artificial intelligence (AI) infrastructure — particularly large-scale data centers — as critical national capacity. Yet such projects frequently encounter resistance, delay, or reversal despite strong technical and economic justification. This article argues that these outcomes are not anomalous implementation failures but predictable consequences of a recurring governance error: the conflation of public policy and public affairs. Drawing on governance theory, policy feedback literature, and systems thinking, the article advances a clear chain of reasoning. First, public policy determines the analytical quality and coherence of infrastructure decisions but does not determine their political durability. Second, public affairs governs legitimacy, stakeholder interpretation, and resistance dynamics once decisions enter the public arena. Third, when legitimacy challenges are misdiagnosed as policy design deficiencies, institutions respond by increasing complexity and delay, thereby activating reinforcing resistance feedback loops. Fourth, early public affairs interventions — particularly stakeholder engagement and trust-building — function as balancing mechanisms that stabilize decisions before escalation becomes path dependent. Using a fictional but empirically grounded case of a national AI infrastructure authority, the article formalizes these dynamics into a feedback model, derives testable hypotheses, specifies empirical methods, and identifies intervention leverage points and boundary conditions. The analysis demonstrates that successful AI infrastructure governance depends less on technical optimization than on correctly sequencing policy design and public affairs intervention within political feedback systems.

Keywords: public policy; public affairs; AI infrastructure; legitimacy; governance; systems thinking; policy feedback

1. Introduction

Artificial intelligence has transitioned from a sector-specific technology to a form of critical national infrastructure, underpinning economic competitiveness, public-sector modernization, and national security strategies across advanced economies (OECD, 2020). As a result, governments have prioritized the construction of large-scale AI data centers as strategic assets. Despite strong analytical foundations — environmental assessments, economic justification, and regulatory compliance — such projects repeatedly face public opposition, political contestation, and procedural delay (Flyvbjerg, 2014).

This article begins from a simple empirical observation: technically sound AI infrastructure policies often fail not at the point of design, but during political and social uptake. Existing explanations typically attribute these failures to communication deficits, local opposition, or political polarization. While each may contribute, they do not explain why analytically similar projects produce divergent outcomes across comparable institutional contexts (Cairney, 2020).

The article advances a different explanation, structured as a causal sequence. First, public policy and public affairs are analytically distinct domains governed by different logics. Public policy concerns the design of authoritative solutions to collective problems; public affairs concerns the political and social conditions under which those decisions are accepted, contested, or resisted (Howlett, 2019; Scharpf, 1999). Second, institutions routinely conflate these domains, assuming that analytically sound policy decisions will translate automatically into public acceptance. Third, this assumption leads institutions to respond to resistance by intensifying analysis and procedural rigor rather than addressing legitimacy dynamics. Fourth, such responses increase perceived complexity and distance, unintentionally amplifying resistance through self-reinforcing feedback loops (Pierson, 1993). Finally, early public affairs interventions — particularly stakeholder engagement and trust-building — can interrupt this escalation by stabilizing legitimacy before resistance becomes path dependent.

By conceptualizing AI infrastructure governance as a system of interacting feedback loops rather than a linear implementation process, this article explains why escalation is predictable, why certain interventions fail, and where high-leverage governance interventions are most effective.

2. Public Policy as Analytical Design

Public policy refers to the structured process through which governments identify collective problems, define objectives, and design authoritative solutions using analytical and normative tools (Dunn, 2018). It addresses what ought to be done based on evidence, institutional mandates, and societal values (Howlett, 2019).

Policy design emphasizes problem definition, evaluation of alternatives, trade-off analysis, and the selection of regulatory or programmatic instruments (Howlett, 2019). In the context of AI data centers, public policy addresses whether compute capacity constitutes essential infrastructure, which environmental and energy constraints should apply, how technological and societal risks should be governed, and how public value is generated relative to cost (OECD, 2020).

Policy coherence is typically assessed internally against criteria of consistency, legality, efficiency, and effectiveness (Dunn, 2018). While necessary, these criteria are insufficient to ensure implementation success in complex political environments (Scharpf, 1999; Flyvbjerg, 2014).

3. Public Affairs as Political and Social Stabilization

Public affairs operates in a distinct analytical domain. Rather than designing solutions, it manages the political and social conditions under which decisions are accepted, contested, or resisted (Scharpf, 1999). Its focus includes stakeholder relationships, power asymmetries, legitimacy construction, and narrative alignment (Suchman, 1995).

From a governance perspective, public affairs addresses a fundamental constraint: decisions do not implement themselves. Acceptance is not a derivative of technical correctness but a precondition for durable authority and compliance (Suchman, 1995). Accordingly, public policy and public affairs answer different but complementary questions: public policy evaluates substantive justification, while public affairs evaluates political and social sustainability (Scharpf, 1999).

4. Conceptual Model: Feedback Governance in AI Infrastructure

AI infrastructure governance is conceptualized here as a dynamic feedback system, consistent with policy feedback theory and systems thinking (Pierson, 1993; Mettler & Soss, 2004). Outcomes are shaped not only by initial design choices but by how decisions generate social and political responses that feed back into institutional behavior over time.

… The full version of this article is available on my Medium publication https://medium.com/@tarifabeach/public-policy-vs-public-affairs-why-confusing-the-two-derails-ai-infrastructure-decisions-cafe96c935c9