In a groundbreaking exploration at the intersection of artificial intelligence and urban planning, researchers have harnessed the power of ChatGPT-4o, an advanced generative AI model, to envision future cities and evaluate their design elements through human-centered analysis. This pioneering study delves deep into how AI-generated imagery can not only inspire but also shape the vital indicators that define the livability and functionality of tomorrow’s urban landscapes. By integrating qualitative expert insights with robust quantitative public surveys, the research provides an unprecedented window into the evolving preferences and priorities of city dwellers, thus setting a new benchmark for AI’s role in shaping sustainable and vibrant urban futures.
The study focuses on eight carefully selected indicators that collectively represent the multifaceted aspirations of future urban environments. These indicators include creativity, traffic rationality, design coherence, environmental greening, public space utilization, technological sense, visual quality, and cultural representation. Each of these dimensions was scrutinized to understand both its perceived importance to residents and its current performance in AI-generated urban designs. The use of Importance-Performance Analysis (IPA) facilitated a nuanced assessment framework, capable of distinguishing indicators that demand urgent enhancement from those that are already well-appreciated.
Creativity emerged resoundingly as the foremost indicator demanding significant improvement. This signals a widespread public yearning for more innovative, bold, and original urban forms that break away from conventional norms. Contrasting this, the indicator of technological sense garnered considerable appreciation for its integration and representation within the AI-generated visions. This duality highlights a critical juncture where cities are both celebrated for their technological sophistication yet simultaneously challenged to foster greater creative ingenuity in their design languages.
Beyond these headline findings, the research reveals subtler dynamics influencing urban design perceptions. While some indicators such as environmental greening and cultural representation were not prioritized as top concerns, their presence in the discourse underscores the ongoing importance of sustainability and heritage preservation within contemporary urban imaginations. Particularly noteworthy is how AI’s ability to creatively augment visual quality offers latent potential that, if fully realized, could revolutionize public engagement with urban architecture and planning.
Methodologically, the study showcases the power of blending focus groups composed of domain experts with broad public surveys to cultivate a holistic understanding of urban design evaluation. The qualitative insights from architects, urban planners, and cultural scholars enriched the indicator selection and contextual framing, while the broad-based quantitative data captured the diverse expectations of everyday city residents. This mixed-methods approach not only strengthens the validity of the findings but also demonstrates how human evaluative frameworks remain indispensable even amid advanced AI-driven methods.
Nevertheless, the research acknowledges inherent constraints tied to employing static images generated by ChatGPT-4o. Static visualizations, while insightful, lack the capacity to depict dynamic, interactive urban phenomena such as social harmony, functional adaptability, and changing environmental conditions over time. These limitations underscore the need for future work to incorporate more immersive and longitudinal modalities like virtual reality or real-time simulations to capture the full spectrum of urban complexity.
Another important consideration is the cultural and regional homogeneity within the expert focus groups, which were primarily drawn from China. This shared background may have subtly influenced the selection of evaluation indicators, potentially limiting the global applicability of the framework. Expanding future research to include experts from a broader range of cultural and geographic contexts will help refine and adapt these indicators to diverse urban experiences, enhancing the universality of AI-assisted urban design evaluation.
The choice to focus exclusively on ChatGPT-4o as the AI model for generating and assessing urban images presents both strengths and shortcomings. This singular focus permits an in-depth exploration of one advanced language model’s creative and evaluative capacities but also narrows the breadth of understanding across the rapidly evolving landscape of generative AI tools. The study advocates for subsequent comparative analyses involving other generative platforms, such as Midjourney and DALL-E, to discern differences in design coherence, cultural adaptability, and aesthetic creativity.
Human involvement in the AI image generation and evaluation process remains an unavoidable factor. While the researchers implemented consistent prompting protocols and limited iterative generations to reduce bias, subjective human judgment was inevitable in selecting final images and identifying generation errors such as image distortions or unrealistic spatial layouts. This underscores the current challenge of balancing human oversight with AI autonomy—a tension that future automation and computational metrics could soon ameliorate.
Looking forward, this study lays vital groundwork for a new research trajectory aimed at democratically blending AI capabilities with human evaluative wisdom in urban planning. Introducing interactive technologies, integrating diverse cultural perspectives, and benchmarking multiple AI models will be key pillars for advancing this interdisciplinary inquiry. Moreover, expanding evaluation metrics to include computational objectivity and broader crowd-sourced assessments will enhance reliability and scalability, making AI-generated urban visions not just imaginative artifacts but actionable blueprints.
What sets this research apart is its holistic vision of AI as not merely a technical tool but as a creative partner capable of reimagining cities through a socially conscious lens. By foregrounding human preferences and societal values throughout the evaluative process, the study envisions an era where AI-generated urban designs can genuinely reflect and anticipate the nuanced demands of future inhabitants. The implications for sustainable development, cultural preservation, and innovative urban aesthetics are profound, suggesting that AI may soon shift from the periphery to the core of city planning practices worldwide.
Furthermore, this research prompts a reevaluation of how urban success is measured in an AI-enhanced world. Traditional metrics grounded solely in functional efficiency might expand to embrace creativity, cultural resonance, and emotional engagement, dimensions where AI can introduce fresh perspectives and challenge entrenched paradigms. By empirically grounding these expanded criteria through resident feedback and expert synthesis, new standards for urban livability and design excellence may emerge.
The study’s emphasis on creativity as a critical area for improvement also reveals a broader challenge facing AI-generated content: balancing novelty with coherence. While AI can rapidly produce imaginative ideas, integrating these ideas into cohesive and context-sensitive urban environments remains complex. This finding carries implications beyond urban planning, touching on the broader AI discourse about how machines and humans collaborate to foster innovation without sacrificing relevance and practicality.
In sum, this investigation into ChatGPT-4o’s role in future city visualization marks a significant milestone in marrying artificial intelligence with human-centered urban design evaluation. By deploying sophisticated analytical techniques and embracing multiple disciplinary viewpoints, the study offers not just descriptive insights but a replicable methodological template for future explorations. Its contributions resonate strongly amid global efforts to imagine more adaptive, inclusive, and aesthetically enriching urban futures in the face of accelerating technological change.
As urban planners, AI developers, and social scientists converge on these innovative frameworks, the path toward smarter, more empathetic cities becomes clearer. This research acts as a beacon illuminating how next-generation AI models can expand the boundaries of urban design imagination, making space for a collaborative, smarter, and deeply human future cityscape.
—
Subject of Research: Application and human evaluation of AI-generated urban design images using ChatGPT-4o, focusing on key urban planning indicators through importance-performance analysis.
Article Title: Future cities imagined by ChatGPT-4o: human evaluation using importance-performance analysis.
Article References:
Cao, Z., Mao, Y., Mustafa, M. et al. Future cities imagined by ChatGPT-4o: human evaluation using importance-performance analysis.
Humanit Soc Sci Commun 12, 630 (2025). https://doi.org/10.1057/s41599-025-04941-6
Image Credits: AI Generated
