Material Considerations

A Normative–Computational Paradigm for Planning Judgement: Integrating Value Pluralism, Consequence Modelling, and AI-Supported Public Reason

Abstract

Planning has always been a site of normative conflict, yet its institutional processes often render those conflicts opaque. Contemporary digital planning efforts frequently oscillate between two poles: optimisation-centric models that reduce planning to algorithmic efficiency, and interpretive traditions that reject any formalisation of judgement. Both positions overlook a crucial possibility: AI can articulate, structure, and illuminate the reasoning environment of planning without determining its outcomes.

This paper proposes a unified normative–computational paradigm—a framework in which AI becomes an epistemic device that (1) makes the public-interest value frame explicit and adjustable; (2) synthesises multi-modal evidence into consequence landscapes; and (3) presents trade-offs through Pareto frontiers that preserve human discretion. Judgement remains a human act, but the conditions under which judgement is made become more legible, contestable, and democratically grounded. The resulting system is neither a decision-making algorithm nor a purely interpretive aid: it is a computational engine of public reason, designed to support transparent and reflexive judgement in a complex institutional field.

1. Introduction: Planning as a Field of Normative Conflict Under Complexity

Planning is not an optimisation problem, nor is it a domain of unconstrained interpretive freedom. It is a site where contested visions of the public interest, embedded in institutional structures, are negotiated through judgement. These normative tensions are typically submerged beneath procedural formality, expressed indirectly through officer reports, committee debates, and appeal decisions. The surface-level neutrality of planning documents often conceals moral architectures that structure decision-making long before any explicit weighing occurs.

Digital planning reform to date has struggled with this normative core. Automation-led approaches treat planning as a computational problem of objective maximisation; interpretive schools insist that judgement cannot be formalised without eroding its integrity. Both perspectives miss the possibility that AI can formalise the conditions of judgement—without formalising judgement itself.

We advance a paradigm in which AI is interwoven throughout planning’s epistemic process. AI does not make decisions: it exposes the structure of value conflict, clarifies evidence, and renders the reasoning space visible. What emerges is a third way: a normative–computational methodology capable of advancing planning theory, supporting practice, and strengthening the democratic legitimacy of decisions.

2. Planning Judgement as Situated Normative Deliberation: The Case for a Computational Support Structure

Planning decisions arise at the intersection of law, policy, place, politics, and uncertainty. Officers and committees navigate:

Judgement in this setting is a form of situated normative deliberation under complexity. Traditional planning theory recognises this but lacks a method to operationalise the environment of judgement without distorting it. AI, as we argue, is uniquely suited to this task—not because it makes decisions, but because it can structure, synthesise, and render navigable the reasoning landscape within which human judgement takes place.

This insight forms the backbone of our normative–computational paradigm.

3. Making Normative Commitments Explicit: Public-Interest Frames as Configurable Inputs

Planning decisions rely on implicit normative assumptions: what counts as harm, who counts as a beneficiary, what constitutes fairness, what futures are desirable. These assumptions are grounded in political priorities and ethical commitments and often reflected implicitly in plan-making and development management.

Our framework formalises this step by allowing planners, councils, or political administrations to articulate—and deliberately modify—their public-interest frame. This frame may draw from:

AI plays a crucial role here. Using LLM reasoning and formal weighting models, the system:

Rather than imposing values, AI supports explicit, reflective, and contestable norm-setting.

4. Moral Cartography: AI-Supported Construction of Consequence Landscapes

Once the normative frame is specified, the system evaluates the spatial and social consequences of alternative options. This is where AI’s integrative capacity becomes essential.

AI synthesises multi-modal evidence by:

Rather than presenting a single deterministic prediction, AI constructs consequence landscapes—interactive representations of how different normative priorities generate different spatial futures. These landscapes reveal distributional impacts, synergies, and externalities that human officers cannot feasibly infer unaided.

The result is moral cartography: computationally mediated insight into how value choices ripple through cities.

5. AI as an Explanatory Engine: Translating Complexity into Legible Reasoning

A key function of AI in this paradigm is not merely analytical but explanatory. Contemporary LLMs excel at generating structured, context-sensitive, human-readable reasoning. When embedded within statutory templates and interpretive fields, AI can:

This moves beyond decision support. AI becomes an engine of public reasoning, making the epistemic structure of judgement visible rather than mystified.

6. Pareto Frontiers as the Moral Horizon of Reasonable Outcomes

Planning problems rarely exhibit a single optimal solution. Under any normative configuration, feasible outcomes form a Pareto frontier: a boundary of non-dominated options that instantiate different trade-off profiles. AI supports this process by:

This redefines what it means for a planning decision to be ‘reasonable’. Reasonableness becomes a matter of selecting from a transparent, evidence-based, normatively structured frontier—not of discovering a singular optimum or writing a justification ex post. AI reveals the horizon of legitimate choice; humans choose which horizon point to inhabit.

7. Structured Interpretive Surfaces: Interfaces for Reflexive and Accountable Judgement

The culmination of this paradigm is the design of structured interpretive surfaces: interfaces that organise AI-mediated reasoning into legally and institutionally appropriate forms. These surfaces:

The aim is not to standardise thought but to enrich the reflective capacity of planners and decision-makers. Structured interpretive surfaces become a mechanism for operationalising public reason within planning practice.

8. Contributions to Planning Theory: A Unified Epistemic Framework

This paradigm offers several innovations:

  1. A fusion of normative theory and computational reasoning, showing that value pluralism can be formalised without eliminating interpretive flexibility.
  2. A reformulation of judgement as the selection of a frontier point within a visible moral and evidential landscape.
  3. A redefinition of AI’s role from decision-maker to epistemic mediator—supporting the reasoning structure without determining the outcome.
  4. A design methodology for interfaces that scaffold public reason, aligning computational capabilities with the institutional logic of planning.
  5. A model for legitimacy, in which normative commitments are explicit, evidence is transparent, and choices are situated within a structured horizon of reasonable alternatives.

Together, these contributions constitute a new epistemic apparatus for planning: neither technocratic optimisation nor open-textured interpretivism, but a computationally enabled practice of normative deliberation.

9. Implications for Governance and Practice

Local authorities, inspectors, policymakers, and communities stand to gain significantly from this paradigm:

This approach aligns with legal expectations of reasonableness and procedural fairness while preserving the core human responsibility of planning judgement.

10. Conclusion: AI as a Device of Public Reason in Planning

Planning does not need algorithms that decide. It needs instruments that illuminate: instruments that clarify how values, evidence, constraints, and consequences shape the horizon of reasonable outcomes. The normative–computational paradigm presented here positions AI as such an instrument—a device that renders the moral architecture of planning visible, contestable, and democratically grounded.

In an era defined by housing crises, climate transitions, and unprecedented urban complexity, planning requires epistemic tools that match the scale of its challenges. A computational public-interest engine is not a replacement for judgement. It is a way of elevating judgement by exposing the reasoning structures that define it.