International Eurasian Academy of Sciences academician and Guangzhou University Professor Zhang Xinchang recently highlighted that while China's smart city construction is progressing well, a significant "last mile" gap remains in areas like space-air-ground integrated data processing and AI application. He believes that as general large models, specialized small models, and high-quality industry data deeply integrate, smart city development will enter a new phase.

Professor Zhang explained that the essence of smart city construction is a complete data value chain. Data is sourced from satellites, drones, and maps, undergoes intelligent perception and effective fusion, builds knowledge graphs and cognitive systems, and ultimately forms decision-making and service models for various industries. He noted that with the arrival of 6G and even 7G, faster and wider space-air-ground data channels will become available. This provides a foundation for improving the precision of technologies like navigation and remote sensing, making the transformation of low-value services into high-value ones an inevitable trend.

However, Professor Zhang pointed out that a wider channel does not guarantee service implementation. "We now face a core contradiction: data acquisition from the sky is getting faster, but the capacity for 'in-space computation' is severely lacking," he said. Addressing the bottleneck in space-air-ground integration, he remarked, "Our ideals are ambitious, but reality is not so." He explained that much satellite data cannot be processed in orbit in real-time and must be transmitted to the ground for calculation, analysis, and service delivery. This creates a significant time lag from data landing to serving society, making it difficult to meet the millisecond-level response demands of traffic control and emergency management. "Hardware efficiency is improving rapidly, but software capabilities and data processing capacity remain weak." He described this as the "last mile" of industrialization, stating, "Whoever can quickly bridge the gap between in-space processing and ground-based services will make smart cities truly 'fast'."

Having researched smart city construction for decades, Professor Zhang believes the overall momentum in China is strong, with progress accelerating. He sees the "first tier" of smart city development primarily in developed regions: Beijing, Shanghai, Guangzhou, and Shenzhen have an early start and significant investment in data infrastructure and IT management. Cities like Wuhan, Suzhou, and Hangzhou are also advancing rapidly. He cited Hangzhou's city brain system, which, when combined with AI, can now achieve adaptive traffic light control. As a benchmark for new city construction, Xiong'an's digital twin concept—creating a city on the ground, a city in the cloud, and a city underground—enables synchronous simulation and interaction between the physical and digital city, treating the city as a living organism.

Regarding the impact of the recent AI boom, Professor Zhang believes AI is bringing revolutionary changes to smart city construction. For urban planning, he noted that the traditional manual process of drawing and revising plans is being upended. Now, teams can simply build a data foundation, input prompts, and automatically generate solutions. He cautioned, however, about the "hallucination" problem of AI in the smart city domain. "General large models primarily learn from internet data, but not all internet data is reliable or accurate." He emphasized that many fields like natural resources, surveying, and mapping contain sensitive data. "These datasets cannot be used for training or iteration, leading to model bias." He suggests that smart city construction needs "smarter large models"—that is, general large models combined with specialized vertical models for fields like natural resources and environmental protection, fused with both public internet data and professional industry data. "Without high-quality, high-credibility professional datasets, even AI will struggle to produce results."

With the low-altitude economy being integrated into national strategy, scenarios like drone logistics and urban air mobility are generating high expectations. Professor Zhang believes the technical foundation for a low-altitude remote sensing network is already in place. However, he noted a major safety issue: "Airplanes have fixed routes and airports, but drones are launched and land freely. If tens of thousands of drones take to the air simultaneously, with intersecting flight paths, safety is a major problem." He argued that for the low-altitude economy to develop, a comprehensive safety network covering the airspace must be established. "We need to transplant ground traffic rules into the sky, creating a unified aerial network, to truly move services like logistics and inspection into the air."

Professor Zhang used the metaphor of "embroidering with precision" to describe the future direction of smart cities. From the digital cities of over 20 years ago, to smart cities, and now to the new generation of smart city construction, technological iteration has never stopped. Yet, the "last mile" of service delivery and the "last meter" of data precision remain the key determinants of success in smart city development.