Symbiotic Planning Theory: The CORE Framework for Human–AI Cocreation in Urban Planning

Problem, research strategy, and findings: Traditional planning theories—rational, participatory, and codesign—lack explicit frameworks for human–artificial intelligence (AI) cocreation at scale. Cities now produce vast real-time information while confronting wicked problems that exceed human analytical capacity, yet existing paradigms offer no guidance for when AI should propose, when humans must authorize, or how to ensure accountability. We developed the symbiotic planning theory (SPT) through theoretical synthesis, and operationalize it via the CORE framework (Collaboration, Options, Refinement, Execution) with embedded safeguards: transparency, fairness constraints, contestation pathways, and time-bound reauthorization. We validated SPT through Gainesville’s (FL) micromobility program, which exemplifies the compliance trap: Vendors meet a 10% equity-zone deployment requirement, yet the equity zone generates only about 3% of trips versus 84% near campus. Our study—the program’s first systematic community engagement in 4 years—identified cost barriers, competing free transit, and payment exclusions. Community contestation imposed binding equity constraints; a deep reinforcement learning (DRL) model was retrained under those guardrails, and continued use was conditioned on outcome-based key performance indicators (KPIs) rather than deployment compliance alone. The framework introduces generative provocation (AI exposes bias in baseline optimization), model pluralism (systematic comparison of alternatives with documented selection rationale), and the equity gate (reauthorization halts when outcome thresholds are breached). Validation relied on a single-city case with nonrandom community inputs and constrained vendor telemetry, limiting causal inference and generalizability. Transferability to housing and climate adaptation is outlined conceptually but requires empirical testing. Framework scalability to resource-constrained agencies depends on interagency collaboration and procurement strategies not yet widely adopted. Takeaway for practice: SPT offers a roadmap for accountable AI integration. Planners remain decision authorities: AI proposes, humans authorize. Planners can apply CORE with concrete artifacts: model cards, audit trails, equity-disaggregated KPIs, public dashboards, grievance workflows, and binding reauthorization tied to equity performance, not just deployment compliance.