Background
As pivotal drivers of smart cities, mega-mobility systems integrate large-scale transportation networks, communication nodes, and energy circuits into a coupled multi-network system. The urban mega-systems epitomize the grand challenge of “organized complexity”, exhibiting characteristic features such as adaptive openness, nonlinear dynamics, hierarchical organization, and emergent properties. Analytical investigations, constrained by rigid separation of macro- and micro-level paradigms, struggle to capture the nonlinear interdependencies across levels that define mega-mobility systems. In this work, the authors systematically advance macro-micro integration with feedback (MMIF) as a transformative paradigm for analyzing urban mega-mobility systems, synthesizing the state-of-the-art developments in typical constituent subsystems under this unified perspective. The MMIF paradigm bridges the gap between theoretical abstraction and empirical practice, contributing to scientifically sound urban development by harmonizing emergent patterns with granular behavioral dynamics. Building upon this paradigm, we investigate the key methods and technologies empowered by artificial intelligence (AI) that enable MMIF, and critically analyze the enduring challenges and prospective research directions. As urban mobility systems increasingly serve as testbeds for complexity science, the MMIF paradigm using AI promises to reshape interdisciplinary collaboration, offering a blueprint for building intelligent, adaptive, and human-centric cities.

Research progress
(1) Paradigm level
A new analytical framework—an AI-empowered Macro–Micro Integration with Feedback (MMIF) paradigm—is systematically proposed. MMIF establishes a “modeling–algorithm–simulation–cross-scale feedback” workflow through continuous, cyclic feedback between macro-level system states and micro-level individual behaviors. Macro states shape local actions, while micro-level changes in turn reshape overall system dynamics, forming an iterative, co-evolutionary modeling cycle. This paradigm provides a theoretical foundation for modeling, inference, and control of coupled transportation–communication–energy multi-network urban systems.

(2) Technical level: four foundational technology pillars
The AI technology stack underpinning MMIF is summarized into four key categories: heterogeneous data fusion, large-scale intelligent computing, knowledge–data collaborative modeling, and generative agents.

1. Multi-source heterogeneous data acquisition and fusion: It integrates sensors, mobile signaling, social footprints, and other streams into unified representations, enabling real-time capture of both macro-level situations and micro-level behaviors and supporting continual model calibration.
2. Large-scale intelligent computing: The nonlinearity and massive scale of urban systems make real-time inference and feedback iterations fundamentally a computing-capacity problem; scalable computing is therefore a necessary condition for operationalizing MMIF.
3. Knowledge–data collaborative modeling: Incorporates mechanistic constraints and prior knowledge to “discipline” data-driven models, mitigating black-box opacity and preventing violations of physical or system constraints.
4. Generative agents: They are used to express reasoning and decision-making in open environments, with potential for causal inference; however, they require careful attention to hallucination risks in complex urban scenarios, reliability under distribution shift, and long-term sustainability in terms of energy consumption.

(3) Application level: coupling across transportation–communication–energy subsystems
Mega-mobility systems are decomposed into three core subsystems—transportation, communication, and energy—and the review systematically synthesizes research progress and application pathways of MMIF across urban transportation (e.g., traffic state estimation and signal control), urban communications (e.g., coverage/capacity optimization and channel allocation), and urban energy systems (e.g., grid load management and charging behaviors). Taking transport–energy interaction as an example, real-time data on individual charging behaviors can feed back to calibrate macro-level demand, while macro-level signals such as electricity pricing can conversely guide individual charging and travel decisions.

Future Prospects
In the era of AI-driven rapid iteration for smart cities, the investigations of urban mega-mobility systems embody a pivotal frontier in 21st-century science. This work establishes the core paradigm of MMIF, which organically combines the emergent properties at the macrolevel with the behavioral mechanisms of microlevel individuals. By introducing continuous, bidirectional feedback between these scales, MMIF transforms macro- and micro perspectives into a self-consistent, co-adaptive modeling cycle, enabling cities to sense, respond, and reorganize dynamically. The authors hope that the MMIF paradigm will inspire researchers and practitioners to collectively redefine the urban systems: not as a collection of problems to be solved but as a complex living entity to be understood, shaped, and coevolved.

The complete study is accessible via DOI:10.34133/research.0982