In recent years, rapid advancements in artificial intelligence (AI) have significantly transformed geographical research methodologies. Large models such as DeepSeek and ChatGPT are catalyzing a shift in geography from the conventional "experience-empirical-simulation" approach to a multi-paradigm framework co-driven by "big data and intelligent learning," offering new perspectives and methods for understanding and interpreting complex geographical issues. In line with this tendency, the human geography community participated in comprehensive discussions regarding the interplay between AI and geography, the transformation of research paradigms, the agency of AI, and its inherent limitations. Several key insights have emerged: AI and geography are mutually empowering, and their deep integration reshapes both knowledge systems and social practices. When using AI, geographers should maintain their scholarly agency in theoretical framing, value orientation, and contextual interpretation, while emphasizing the situated meaning of human-environment systems and the practical utility of knowledge. This approach fosters a new disciplinary paradigm characterized by "human-machine-environment" synergy. Furthermore, although AI, as a non-human agent, is increasingly involved in the production of geographical knowledge (for example, the concept of a "digital sense of place"), understanding the complexity of human-environment relationships, interpreting socio-spatial dynamics, and appreciating and preserving local experiences must remain the prerogative of geographers, and cannot be supplanted by AI.
The rapid development of Artificial Intelligence (AI) has enhanced the teaching efficiency of geography education and broadened the channels of knowledge dissemination. It has also profoundly impacted traditional teaching models, assessment systems, and talent cultivation objectives. To address this challenge,this study integrates the teaching practices and research insights of scholars from multiple universities. It systematically analyzes the in-depth impacts of AI on geography education and its unique disciplinary characteristics, explores AI-driven transformation paths, and summarizes the core consensus as follows. First, geography education, which integrates the rigor of natural science with humanistic values, is entering a critical period of transformation driven by AI. Although AI can be leveraged to improve teaching efficiency, expand practical scenarios, and optimize personalized teaching, it is essential to clarify the instrumental role of AI and avoid the risks caused such as overreliance, the erosion of students' skills, diminished critical thinking, and ethical concerns. Second, the core competitiveness of geography education lies in spatial thinking, place perception, dialectical analysis, and humanistic spirit—none of which AI can replace. The key to transformation is to adopt the new model of "technology empowerment + competence orientation + integration of virtual and real practice." This approach strengthen students' understanding of natural laws and practical operation capabilities, cultivate their systematic thinking and empirical literacy, enhance their humanistic qualities, enable geography to solidify its roots while embracing frontier technologies. Third, geography educators must transform from knowledge transmitters into mentors and educational practitioners. By redesigning the curriculum system and reforming the teaching evaluation mechanism, they can guide students from "being able to use AI" to "being good at using AI," cultivating compound geography talents with technical literacy, humanistic awareness, spatial thinking, and innovative capabilities.
With the pervasive penetration of artificial intelligence (AI) technologies, traditional paradigms of social space governance are undergoing a fundamental shift—from "digital governance" to "intelligent governance." In the governance space dimension, AI innovations such as AI-generated content (AIGC) and spatial intelligence have endowed digital twin spaces with unprecedented capabilities, transforming them from static reflections of physical reality into dynamic systems capable of proactive inference, simulation, and real-time optimization. This transition extends governance functions beyond mere representation to include predictive intervention and anticipatory regulation. On the dimension of governance subjects, algorithms have qualitatively mutated—from passive instruments of execution into "artificial agents" or "auxiliary governance actors" possessing autonomous learning, environmental adaptation, and predictive decision-making capacities. This mutation fosters an emergent symbiotic "human-machine collaboration," challenging established power structures and reconfiguring accountability boundaries. In this study, building upon this analysis of the data-to-intelligence governance transition, we examine—through case studies including Shanghai's "Quantum City" and Hangzhou's "City Brain"—the expansion logic and practical manifestations of multidimensional social space governance in the AI era. First, the governance space has expanded from a tripartite "physical-social-data" framework to a quadrilateral "physical-social-data-algorithm" structure. Spatial intelligence technologies and "real-world model" paradigms have positioned algorithms as the core of digital twin systems. Empowered by spatial intelligence, digital twin environments achieve heightened precision and synchronicity, enabling real-time and efficient interactions with physical spaces while demonstrating enhanced generative capacity and operational autonomy. These developments constitute the multidimensional spatial arena of public governance. Second, the governance subject has evolved from a "government-market-society" triadic relationship to a "government-market-society-intelligence" quadrilateral synergy. As AI agents gain greater autonomy, their subjectivity becomes increasingly manifested, elevating AI from a mere instrument to a co-constitutive governance actor that must operate in parallel with traditional subjects. This transformation necessitates fundamental theoretical and practical interpretations of the relationships among all governance stakeholders. This multidimensional expansion has engendered a series of novel challenges for public governance practices. First, AI and digital twin technologies have accelerated the convergence of the physical, social, and digital domains, yet this nascent "hybrid space" has precipitated profound normative conflicts in governance practices. Second, as AI transitions from a tool to an intelligent agent, algorithmic bias becomes more acute, and an "accountability vacuum" risk emerges within human-machine collaborative frameworks. Finally, persistent digital divides are metamorphosing into a new configuration—the "intelligence divide"—exacerbating social stratification. To address these emergent challenges, social space governance in the intelligence era requires innovative pathways. First, cross-spatial coordinative governance mechanisms must be constructed to enable the synergistic integration of virtual and physical domains, shifting from normative fragmentation to spatial order reconstruction. Second, a human-machine coordinative governance framework should be built upon technical foundations of "trustworthy AI" and institutional safeguards ensuring "ultimate human control." Third, governance must uphold a people-centered value orientation, ensuring that the benefits of intelligent governance are equally distributed across all citizens.
Spatial Intelligence (SI) is the cornerstone of Artificial Intelligence (AI) development and represents the integrated capability of perceiving, reasoning, and acting within three-dimensional environments. Despite its significance, the geographical community are yet to systematically explore the operational mechanisms of spatial intelligence and its social impacts. Existing research primarily focuses on technological aspects such as digital twins and spatial heterogeneity modeling, while overlooking the profound social transformations that accompany the deployment of SI. As SI applications permeate autonomous driving, embodied robotics, and smart city infrastructure, fundamentally reshaping human-land interaction patterns, this research gap has become increasingly critical. In this study, we employ a cross-disciplinary literature synthesis approach, integrating perspectives from geography, computer science, and social theory to construct a comprehensive analytical framework for examining technological evolution trajectories and their societal impacts. The core objective is to systematically elucidate multidimensional developmental process of SI and reveal its concomitant social restructuring effects. Through a critical analysis of cutting-edge research and empirical cases, we explore how SI evolution fosters novel spatial practices while triggering structural societal challenges. The methodology focuses on integrating literature themes centred around three core capabilities of SI, supplemented by a socio-theoretical analysis of unintended consequences. The study findings reveals three key technological transformations. First, spatial perception has transcended one-dimensional static representation to achieve three-dimensional dynamic understanding. This shift encompasses a transition in representation from linear encoding to voxel/point-cloud-based 3D modeling, a shift in reference frameworks from absolute coordinate systems to dynamic context-aware systems, and a change in cognitive units from isolated objects to spatiotemporal events. Second, spatial reasoning evolved from deterministic rule systems to probabilistic generative models. This transformation includes cognitive mechanisms shifting from formal logic to probabilistic prediction, learning paradigms evolving from supervised training to world-model-based reinforcement learning, and expression forms upgrading from abstract symbolic descriptions to multimodal embodied interactions. Third, spatial action has transcended the stage of situational adaptation and is advancing toward spatial co-creation. This phase is characterized by: the diversification of agents, where human actors collaborate with increasingly autonomous AI actors in shared environments; and a shift from unidirectional reception to bidirectional co-construction in interaction modes, epitomized by the "Industry 5.0" paradigm emphasizing on proactive human-machine collaboration and natural interaction interfaces. However, these technological transformations have generated significant social restructuring. The digital divide is exacerbated by multiple accessibility and usability barriers. Intelligent infrastructure's reliance on high-performance computing widens regional disparities, while the required technical literacy creates an application gap, disproportionately affecting developing regions and marginalized groups. Concurrently, privacy concerns intensify as intelligent infrastructure conducts a massive-scale collection of spatial, behavioral, and biometric data. Furthermore, legal frameworks lag significantly behind the rapid development of smart infrastructure. Defining liability within complex human-machine-human interaction networks proves challenging, and emerging rights issues, such as virtual property and algorithmic agency, remain unresolved, as evidenced by protracted litigation over autonomous vehicle accidents. In summary, we posit that smart infrastructure development faces a dual imperative: enhancing technical capabilities and proactively addressing socio-ethical challenges. We propose a responsive intelligent infrastructure framework that integrates value-sensitive design with contextual ethical reasoning and embeds geoethics and spatial justice as core design principles. Future development should prioritize interdisciplinary integration with psychology and sociology, shifting research from "technology-driven" to "problem-driven" approaches, and developing novel architectural systems capable of managing complex, multiscale social ecosystems. This study contributes on three levels: theoretically, it systematically analyzes the social effects of the intelligent society within geographical discourse for the first time; methodologically, it integrates interdisciplinary perspectives to bridge technical and social analysis; practically, it provides actionable insights for policymakers to harness the inclusive potential of intelligent society while mitigating risks, thereby, advancing the "AI for Society" agenda and offering theoretical guidance for intelligent society development.
Artificial intelligence (AI) has profoundly reshaped human society and significantly impacted academic research. In the current era of intelligence, geography requires the development of new theoretical frameworks. By constructing and elaborating the theoretical framework of the "Poetics of Life," this study explores new pathways for geographical expression within the context of human-AI integration. The relationship between AI and humans manifests in four modes: tool, partner, friend, and soul. In the process of human-AI integration, place, body, and emotion emerge as three key vectors that are currently irreplaceable by AI. Drawing on existentialist philosophy, geographical poetics, and humanistic geography, and employing a digital autoethnographic approach, this article conducts an in-depth analysis of 122 songs co-created by Ye Chao(The first author) and AI and published on the personal WeChat Channel "Ye Shenxun." It compares the characteristics of individual writing with those of human-AI collaborative creation and summarizes their public communication effects. What distinguishes Poetics of Life in the new era from geographical poetics lies in three fundamental shifts: the creative subject has transformed from a solitary author to human-AI co-creation, the form of expression has expanded from single-text delivery to multisensory stimulation, and media dissemination has evolved from one-way output to multidimensional interaction. The song samples exhibited diverse styles and themes, reflecting the interplay of emotion, place, and AI, thereby highlighting the importance of new forms of geographical writing and expression in the intelligent age. In terms of communicative effects, a top-ten analysis of the texts revealed that audiences with a background in geography paid more attention to the mutual construction of place and everyday life, whereas other audiences focused more on emotional resonance. Surreal works, such as Chronicle of Light and Dust, demonstrate a cross-disciplinary, future-oriented dimension. The Poetics of Life in the intelligent age not only extends and deepens the humanistic tradition of geography but also provides new theoretical insights for interdisciplinary fields such as digital art and media geography. The expression, performance, and public communication of the Poetics of Life constitute key directions for future research.
In recent years, advances in pre-training techniques and improvements in computing hardware have led to substantial breakthroughs in Artificial Intelligence (AI). Large-scale models, such as ChatGPT and DeepSeek, have demonstrated unprecedented capabilities in natural language processing and generation, accelerating the deployment of intelligent technologies across diverse domains. Nevertheless, current large models still face notable challenges in terms of physical common-sense understanding, causal reasoning, and the modeling of dynamic environments. In response to these deficiencies, the concept of "World Models" has recently emerged, with the goal of constructing cognitive engines that internally model, simulate, and predict physical environments. In this review article we describe the origins and research pathways of World Models, tracing their technical evolution from representation learning and dynamic modeling to embodied interaction. We summarize the core approaches to understanding environmental structure, simulating future states, and supporting decision-making and reasoning. From a geographical perspective, the generative, multimodal, and interactive capabilities emphasized by World Models are regarded as key requirements for characterizing complex spatial structures and dynamic processes. These capabilities are conceptually aligned with key research topics in geography: spatial organization, behavioral processes, and interactions with the environment. With the development of video generation, large-scale multimodal learning, and embodied intelligence, the field of AI is increasingly shifting from symbolic descriptions of the world to computable forms of spatial cognition, reflecting an intelligence paradigm fundamentally oriented toward space. The advancement of World Models not only provides new ways in which AI can understand the structure and processes of the real world, but also offers important opportunities for geography to explore spatiotemporal process modeling, mechanisms of spatial cognition, and the construction of integrated virtual-physical environments. With this overview we seek to establish a systematic framework for interdisciplinary research at the intersection of AI and geographical science and to provide references for future studies on spatial intelligence and AI.
The advent and deep permeation of digital media technologies have precipitated a paradigmatic shift in the ontological and phenomenological understanding of place. No longer conceived as a stable, bounded physical container, place in the contemporary era is dynamically produced, mediated, and continually reconstituted through the intricate interplay of platform architectures, algorithmic operations, locative media, and networked social practices. This transformation has catalyzed the emergence of "digital sense of place" as a critical interdisciplinary concern. Scholars across the disciplines of human geography, environmental psychology, sociology, anthropology, and media studies have engaged with this phenomenon, yet their inquiries have largely progressed in parallel, resulting in a fragmented intellectual landscape characterized by conceptual dispersion, methodological insularity, and theoretical compartmentalization. A cohesive framework capable of elucidating the complex, recursive coupling between the digital and the placal remains conspicuously absent. To address this gap, this article conducts a systematic knowledge archaeology and synthesis of relevant literature spanning the years 1980 to 2025. Employing bibliometric analysis and critical discourse review, we trace the genealogical development of place scholarship within each of the core disciplines and map their convergent trajectories toward the digital. Our analysis identifies a fundamental theoretical evolution: a move from essentialist, static, and physically deterministic models of place (exemplified by Tuan's topophilia and Relph's place identity) toward relational, processual, and mediated conceptions. Human geography's "relational turn" and its subsequent engagement with "hybrid space" dismantled the physical-digital binary. Environmental psychology meticulously operationalized and measured place attachment, later extending its quantitative paradigms to validate the psychological reality of digital emotional bonds. Sociology and anthropology foregrounded the social construction of place, revealing how power dynamics, cultural practices, and embodied rituals undergird place-making—a perspective extended to digital communities and virtual belonging. Media studies evolved from treating media as mere representational tools to recognizing platforms and locative media as constitutive infrastructures that actively shape spatial perception and social interaction. The synthesis of these multidisciplinary insights exposes their collective yet unintegrated recognition of digital sense of place as a multifaceted, systemic phenomenon. Building on this foundation, this paper makes a central theoretical contribution by proposing "digital sense of place" as a systemically generative integrative analytical framework. This framework posits digital sense of place not as a possessed attribute but as an ongoing, emergent process generated within a dynamic system composed of five interconnected subsystems: (1) the technological-infrastructural subsystem (platforms, algorithms, interfaces); (2) the affective-psychological subsystem (digitally mediated attachment, identity, meaning); (3) the social-relational subsystem (networked communities and mediated interactions); (4) the cultural-semiotic subsystem (the remediation and circulation of place-based narratives and memories); and (5) the power-political economic subsystem (the governance, ownership, and algorithmic curation of digital space). These subsystems operate in continuous feedback loops, co-constituting the lived experience of place in a digital society. This systemic, generative perspective facilitates a critical analysis of core tensions inherent in digital place-making, such as between delocalization and re-localization, authentic affective experience and platform-engineered engagement, and discursive openness and algorithmic exclusion. Consequently, this integrated framework advances the field from multidisciplinary parallelism toward theoretically robust, holistic explanation. It provides a potent lens for examining pressing contemporary issues, including the affective politics of platform societies, the governance of smart cities, the preservation of digital heritage, and the ethical implications of algorithmically modulated spatial experience. The framework thus repositions digital sense of place as a central analytical node for understanding how locality is persistently forged, contested, and lived within the matrix of contemporary techno-social life.
In the context of the digital age, digital humanities has evolved into a cutting-edge interdisciplinary field that deeply integrates computational and humanistic thinking, representing a fundamental shift in research paradigms driven by data, employing computation as a methodological approach, while remaining guided by the humanities. The importance of spatial representation has become increasingly prominent in this paradigm shift. Visualization technology is particularly crucial compared to other spatial representation methods. As a core method in digital humanities research, it can closely match computational models to humanities data, thereby building a bridge for interdisciplinary research. This study aims to systematically analyze the design principles and implementation paths of typical visualization technologies based on a literature review and to reveal their methodological value in the cutting-edge field of digital humanities. Spatial thinking in human geography underwent four major shifts. This study proposes a new framework for three visualization paths based on the spatial production theory, arguing that spatial representation from a digital humanities perspective is a digital translation of the geographical spatial triad. First, text visualization corresponds to "conceptual space," transforming massive amounts of unstructured text into abstract cognitive maps, aiming to reveal the power logic and cultural imagination behind discourse. Second, geographic visualization corresponds to "experiential space," using geographic information technology to locate data in concrete perceptual maps, aiming to reconstruct the locality of historical events and humanistic activities. Third, relational visualization corresponds to "lived space." Through nongeographical networks of actors or topologies of power and capital, it aims to reflect the reshaping of geographical patterns through social relations. The dialectical relationship among these three elements constitutes a methodological innovation in the expression of the digital humanities space, manifested as a three-dimensional model that is cyclically reinforcing and ultimately unified in specific humanities issues. These new visualization paths help to grasp the multifaceted nature of the concept of "space" in digital humanities, covering both mainstream technological methods and connecting the key directions of humanities research—meaning, place, and connection. They also aim to transcend instrumental technological applications, focusing on how to reconstruct the "spatial imagination" of humanities research through digital methods: methodologically, transforming humanities objects into computable spatial models; epistemologically, revealing hidden spatial logic through visual representation; and in terms of discipline construction, providing a clear research paradigm for digital humanities. In short, the triadic framework of text, geographic, and relational visualization systematically responds to the methodological need for digital humanities to provide an operational and computable interpretation of "space." It precisely corresponds to the complete spectrum of humanistic spaces—conceptual space, experiential space, and lived space—and by promoting cross-validation and fusion analysis of multi-dimensional data, propels digital humanities research from fragmented data presentation to holistic and explanatory spatial integration and theoretical generation. Notably, in this process, we must be wary of both technological instrumentalism and visual centrism, thereby providing a positive reference for the paradigm shift and discipline construction of digital humanities. In particular, new paths for spatial representation must establish a balance between computational precision and humanistic depth, ultimately promoting a shift in research paradigms from purely technology-driven to human-centered intelligent reconstructions.
Against the strategic backdrop of "Digital-China" and the "Dual-Carbon" goals, the synergistic advancement of digital economy and carbon emission reduction is crucial for achieving high-quality, sustainable development. As a leading region in China's economic and digital transformation, the Yangtze River Delta (YRD) urban agglomeration provides a critical-case study for examining the complex interplay between digital growth and decarbonization. In this study, we aimed to systematically analyze the spatiotemporal-coupling characteristics and underlying influence mechanisms between the digital economy and carbon emissions in the YRD region from 2011 to 2023. Moving beyond aggregate-analysis and linear-assumptions, this study seeks to reveal the spatial heterogeneity, nonlinear-relationships, and threshold-effects to provide a nuanced empirical basis for differentiated-regional policymaking. Methodologically, we integrated the Geographically Weighted Random Forest (GWRF) model with SHapley Additive exPlanations (SHAP). We constructed comprehensive evaluation systems for both the digital economy and carbon emissions, and calculates the coupling coordination degree (D) between these two systems for 41 cities. The core analytical approach uses the GWRF model, which embeds a spatial-weight matrix into the Random Forest algorithm to simulate the spatially-varying and nonlinear effects of multiple influencing factors on the degree of coordination. Subsequently, the SHAP framework was applied to interpret the GWRF " black-box model and quantify the global-importance, directional-contribution, and potential nonlinear or threshold-behavior of each explanatory variable. This study yielded several key findings. Regarding temporal evolution, the overall coupling coordination degree of the YRD urban agglomeration shows a clear upward trend, increasing from 0.411 in 2011 to 0.505 in 2023, marking a transition from an "imminent-imbalance" to a "barely-coordinated" stage. However, this progression is not monotonic; the significant dip observed in 2021 reflects dynamic tension and potential lagged-adaptation between technological-advancement cycles and stringent emission-reduction targets. In terms of spatial patterns, a distinct hierarchical "core-corridor-periphery" radial structure has formed. Shanghai, leveraging its advanced technological foundation and institutional advantages, remains at the forefront, achieving "high-quality coordination" by 2023. The provinces of Jiangsu and Zhejiang exhibit follow-up growth, entering the "barely-coordinated" stage. In contrast, Anhui province, despite exhibiting the fastest growth rate, remains at the threshold of "imminent-imbalance," highlighting persistent regional disparities within the agglomeration. At the city level, high-coordination cores were concentrated along the Shanghai-Nanjing-Hefei-Hangzhou development axis, with coordination levels gradually diffusing along major transport corridors and weakening in northern Anhui and southwestern Zhejiang. Concerning the model validation and identification of key drivers, the GWRF model demonstrated significantly superior explanatory power and predictive accuracy compared to the standard-Random Forest model, confirming its efficacy in capturing spatial-non-stationarity. The SHAP analysis identified variables from the digital economy subsystem, specifically, the number of mobile phone subscribers, employees in information transmission and software services, and postal business volume, as important positive drivers. Their intensity-of-influence exhibited a spatial-diffusion pattern, radiating outward from core metropolitan areas to key manufacturing nodes and emerging industrial zones. Conversely, variables from the carbon emissions subsystem, particularly carbon emissions intensity and per-capita carbon emissions, act as primary inhibitors of coupling coordination. In summary, this study elucidates a dual-path mechanism, wherein the agglomeration of digital elements drives synergistic improvements, whereas high-carbon economic structures exert inhibitory pressure. This study makes substantive contributions to both the theoretical and methodological fronts. Theoretically, it provides robust empirical evidence for the complex, nonlinear-interdependencies between digital and green transitions, challenging simplistic linear-assumptions and enriching the understanding of their coupling dynamics in a regional context. Methodologically, the integrated GWRF-SHAP framework was validated as a powerful tool for dissecting high-dimensional and spatially-heterogeneous problems in urban and regional studies, offering a replicable-analytical pathway. These findings provide actionable-insights for policymakers to advocate tailored-strategies that reinforce positive digital diffusion, especially in lagging areas, while implementing targeted measures to decouple economic growth from carbon emissions in high-pressure zones. Ultimately, this approach aims to foster a more balanced and synergistic development pathway for the YRD and similar regions.
This study, grounded in the symbiotic relationship between art and geography, constructs a three-dimensional analytical framework composed of data media, algorithmic generation, and multisensory perception. The framework aims to investigate how digital technologies reshape modes of spatial production, sensory engagement, and environmental imagination in contemporary art geography. In the digital era, where data, algorithms, and embodied experience intersect, artistic practices have evolved from static spatial representation to dynamic, participatory, and process-oriented forms of geographic expression. By examining six representative digital artworks including Machine Hallucinations: Nature Dreams and Quantum Memories by Refik Anadol, Monte by Luciano Piccilli, The Deep Listener by Jakob Kudsk Steensen, Peace can be Realized Even without Order by teamLab, and I'm Feeling Lucky by Timothy Thomasson, this study delineates three major thematic dimensions of digital art geography: urban visual documentation, embodied environmental expression, and artificial landscape reconstruction. The analysis reveals that contemporary digital art transforms geographical data into creative material media, using algorithms to generate fluid, dynamic spatial representations that transcend traditional boundaries between the physical and the virtual. Through multisensory and immersive interactions, these artworks invite audiences to inhabit hybrid environments in which data flows, algorithmic processes, and bodily perception are intertwined. Such practices articulate a new form of spatial consciousness, in which human beings experience themselves as simultaneously embedded in both natural and digital ecologies. The study identifies three distinctive tendencies: the transition from artificial selection to data-driven narrative in urban visualization, the shift from philosophical abstraction to informational reconstruction in environmental expression, and the evolution from static fixation to responsive symbiosis in artificial landscape creation. Beyond formal innovation, these transformations embody deeper epistemological and political implications. Digital art geography reveals how data infrastructures, algorithmic systems, and technological mediation carry implicit forms of power that shape perception and representation. The artworks analyzed demonstrate that data are never neutral; rather, they participate in constructing spatial hierarchies and determining who can access, interpret, or intervene in digital landscapes. This study thus extends the critical scope of art geography by engaging with pressing issues of the artificial intelligence era, such as data sovereignty, algorithmic transparency, and the ethical tension between human-centered perception and ecological complexity. In conclusion, digital art geography represents not merely a technological advancement in artistic media but also a profound transformation in the human relationship with space, environment, and information. By integrating artistic creativity with geographic inquiry, the "data medium-algorithmic generation-multisensory perception" framework provides an effective methodological tool for analyzing how digital art redefines the production of space and the aesthetics of inhabitation. Simultaneously, new theoretical and ethical questions concerning how human beings negotiate agency, embodiment, and coexistence within an increasingly algorithmic world are raised. This research contributes both to the theoretical development of art geography and to broader debates on perception, ecology, and critical spatial practice in the digital age.
In the era of Artificial Intelligence (AI), Large Language Models (LLMs) have become important informational mediators through which the public perceives urban spaces, and AI discourse has emerged as a powerful force in the construction of urban spaces. Using Guangzhou as a case study, we categorized urban consumption spaces into four types: shopping spaces, catering and entertainment spaces, tourism and leisure spaces, and residential and commercial housing. We constructed an evaluation question set for hallucinations in urban consumption spaces within a discourse-power framework and used hallucination tests to examine the commonalities and differences between Chinese and international AI models, namely, DeepSeek and ChatGPT, in the production of spatial discourse, thereby explaining how AI hallucination discourse constructs urban consumption spaces. The main findings of this study are as follows. (1) In the hallucination tests of urban consumption spaces, ChatGPT exhibited lower hallucination rates than DeepSeek at both the overall level and across individual categories. Residential and commercial housing emerged as high-incidence domains of hallucinations for both AI models, whereas the most pronounced divergence in hallucination rates between the two models occurred in tourism and leisure spaces. The primary sources of AI-generated content, in descending order, were news media, individual or commercial institutions, government agencies, and online encyclopedias. Both models tend to respond to mainstream spatial discourses, demonstrating a limited capacity for revealing the complex, diverse, and contradictory realities of the city. Specifically, ChatGPT favors generalized frameworks in its depiction of urban consumption spaces, whereas DeepSeek's spatial narratives display a planning-oriented logic aligned with urban development strategies. (2) By integrating and reproducing specific discourses originating from governments, news media, and commercial institutions, AI discourse operates as a novel power subject that constructs multiple "realities" and promotes the production of meanings attached to consumption centers, symbolization of architectural landscapes, and technologization of consumption spaces and also adjudicates spatial value, allowing its power to operate in a "rational" manner. (3) The AI hallucination discourse constructs space by producing subject positions tailored to users, such as "supporters of urban development," "experience-oriented consumers," "beneficiaries of technological progress," and "astute investors." As users identify with and accept these positions, they enact specific consumption-space practices grounded in particular forms of knowledge, generating new data that are subsequently mobilized to reproduce the same discursive system. In this process, a specific knowledge regime is sustained, and power continues to operate. From a discourse-power perspective, this study elucidates the pathways through which urban consumption spaces are constructed by AI in the era of artificial intelligence. Although, it advances our understanding of the modes and impacts of urban knowledge circulation amid the rise of generative AI, critical reflection on the discursive and power relations embedded in technological products contributes to ethical scrutiny of smart city practices.
Artificial intelligence (AI) has emerged as a key driver of high-quality regional development by reshaping innovation systems, industrial structures, and spatial economic dynamics. Consequently, the scientific measurement of the spatial distribution and evolutionary trajectories of AI technologies has become a critical issue in economic geography. Existing empirical studies typically measure AI activity using enterprise registration data or granted invention patents based on proxy variables, keyword searches, or the International Patent Classification system. However, these methods often suffer from limited semantic accuracy and incomplete coverage, making it difficult to fully capture the rapidly evolving and context-dependent nature of AI technologies. To address these limitations, this study developed a semantic-based identification framework based on large language models. Drawing on approximately 1.2 million granted invention patent abstracts from Guangdong Province between 2001 and 2021, we employed Bidirectional Encoder Representations from Transformers (BERT) large language model to identify AI-related technologies based on deep semantic understanding. This approach yielded a dataset of approximately 200,000 AI-related patents and provided a more comprehensive and accurate representation of regional AI innovation activities. Building on this dataset, we applied BERTopic for topic modeling to identify major technological themes and trace their temporal evolution. The empirical results reveal several key findings. (1) From a temporal perspective, the evolution of AI technologies in Guangdong Province followed a clear two-stage trajectory. During the initial stage from 2001 to 2014, AI patenting activities remained at a relatively low level, gradually increasing from 37 patents in 2001 to 3,514 in 2014. By contrast, the period from 2015 to 2021 represents a phase of rapid expansion, characterized by a sharp increase in AI patenting activities and a substantial acceleration in innovation intensity. This shift indicates the growing strategic importance of AI in regional innovation systems. (2) From a spatial perspective, AI technologies are highly unevenly distributed across Guangdong Province, exhibiting strong agglomeration in the Guangdong-Hong Kong-Macao Greater Bay Area. Shenzhen and Guangzhou together account for 75.1% of all AI patents in the province, forming a pronounced core region of AI innovation. Shenzhen contributed to more than half of the provinces' AI patents, demonstrating a strong primacy position. Beyond these two leading cities, Dongguan, Zhuhai, and Foshan constituted the secondary tier in terms of patent volume. Further analysis of co-invention patents revealed the network characteristics of AI technological collaboration. Within Guangdong Province, inter-city cooperation exhibited a clear dual-core structure centered on Guangzhou and Shenzhen, with dense collaborative linkages concentrated in the Greater Bay Area. While Shenzhen dominates AI patent production, Guangzhou demonstrates the highest level of intraprovincial collaboration, indicating a stronger coordinating and connective role within regional innovation networks. (3) In terms of technological content, topic modeling identified five major AI technology themes: data and image processing, robotics and automation devices, intelligent transportation and fault detection, smart homes and environmental control, and bio-simulation and image analysis. Among these themes, data and image processing constituted the most active and foundational domains throughout the study period, entering a phase of rapid growth around 2013 and peaking in 2019. Robotics, intelligent transportation, and smart home technologies have expanded markedly after 2015, reflecting the increasing diversification and application-oriented nature of AI innovation. By contrast, biosimulation and image analysis exhibited modest growth, suggesting a narrower range of applications. Moreover, cities within Guangdong displayed differentiated thematic advantages, reflecting the distinct trajectories of regional AI specialization. Shenzhen has maintained a leading position in image and data processing, as well as robotics; Guangzhou has developed distinctive strengths in intelligent transportation and urban service applications; Zhuhai integrated AI into its home appliance manufacturing base and marine technologies; Dongguan focused on AI applications in intelligent manufacturing and environmental governance; and Foshan emphasized the integration of smart home technologies with industrial automation.
Advancing digital-twin city development through new infrastructure and promoting the upgrading and transformation of smart cities is critical for building new competitive advantages for cities and is an inevitable requirement for modernizing urban governance systems and capabilities. Since new infrastructure and digital-twin cities share similar construction goals, this study examines 149 smart city pilot projects in China from 2011 to 2022, treating new infrastructure as a critical path to the construction of digital twin cities. First, a three-stage data envelopment analysis was used to assess the potential of digital twin city construction from the perspective of new infrastructure. Subsequently, a multi-stage traditional Markov model and a multi-stage spatial Markov model were used to study the potential of digital twin city construction from the perspectives of temporal evolution and spatial spillover, respectively. The results show that: 1) In the first stage, the nationwide overall technical efficiency generally maintained an upward trend, with the efficiency values in the Northeast, West, and Central regions showing a fluctuating upward trend, whereas the overall technical efficiency value in the eastern region remained at a low level. 2) In the third stage, after removing environmental factors and random disturbances, high-level areas were concentrated mainly in the Beijing-Tianjin-Hebei urban agglomeration, eastern Shandong, and some provincial capitals. Further comparison of the comprehensive technical efficiency values between the first and third stages revealed that the eastern region had a higher efficiency than the central region, indicating greater potential for digital twin urban construction. In contrast, the efficiency values of the northeastern, western, and central regions declined. 3) Traditional Markov models show that digital twin urban construction potential has a significant path-dependent effect, and the upgrading trend of digital twin urban construction potential gradually strengthens with an increase in the number of phases. Further introduction of spatial Markov models, accounting for the spatial correlation between neighboring regions, revealed that the evolution of digital twin urban construction potential was substantially affected by the development levels from neighboring areas, with spatial spillover and siphon effects coexisting. Simultaneously, the impact of neighboring cities at different levels of development on the potential evolution of their respective regions also showed significant differences. The contributions of this study are as follows: 1) From a research perspective, it regards the construction of new infrastructure as an important supporting path for the construction of digital twin cities, constructs an analytical framework for new infrastructure to empower the construction of digital twin cities, and measures the potential of digital twin city construction from the perspective of new infrastructure, thus expanding the empirical dimension of cross-disciplinary research on new infrastructure and digital twin cities. 2) In terms of indicator construction, it systematically constructs an indicator system covering the industrial space, social space, governance space, and information space of digital twins, more comprehensively depicting the intrinsic characteristics and development potential of digital twin cities. 3) In terms of research methodology, this study adopts a three-stage Data Envelopment Analysis model to effectively eliminate the influence of environmental factors and random disturbances on efficiency measurement, combining multi-period traditional Markov models and spatial Markov models to reveal the path dependence, gradient evolution, and evolutionary mechanism of spatial spillover and siphon coexistence of the potential of digital twin city construction from the dual dimensions of temporal evolution and spatial correlation.
The preservation of traditional architecture, which is a tangible carrier of regional culture and identity, requires precise and scalable methods for style classification. Here, we address the critical need for such methodologies by developing and applying an integrated computational framework that synergizes deep learning with spatial analytical techniques from cultural geography to document and analyze traditional architectural heritage. The primary research objectives were threefold: first, to construct a high-performance automated system for quantitatively classifying major Chinese traditional architectural styles(CTASs) from visual data; second, to leverage the system's outputs to analyze the multiscale spatial patterns of these styles; and third, to interpret these patterns to identify cultural boundaries and regional diversity, thereby providing a data-driven foundation for heritage conservation and planning. To achieve these aims, the research methodology began with the curation of a multimodal image database derived from 646 nationally designated traditional villages. An EfficientNet model, enhanced for multiscale feature fusion to capture both global and local details, was trained on this dataset to classify six typical styles: Jing, Jin, Chuan, Wan, Su, and Min. To ensure transparency and interpretability, Grad-CAM was used to visualize the key architectural components that informed the model's decisions. Subsequently, the geographically referenced classification results were subjected to a comprehensive spatial analysis suite, including spatial autocorrelation, standard deviational ellipse analysis, and calculation of Shannon diversity and Simpson evenness indices at the provincial level. The results and conclusions of the integrated analysis are detailed and multifaceted. The EfficientNet model achieved an overall classification accuracy of 90%, confirming the efficacy of the deep-learning approach. However, a misclassification rate of 10.7% was observed between the Jing and Chuan styles. Grad-CAM analysis provided critical insights into this phenomenon, revealing that the model's confusion stemmed from a shared, significant focus on similar wooden walls and colonnade features in both the Jing and Chuan styles, highlighting subtle inter-style visual overlaps. At the macroscale, the spatial distribution analysis confirmed that CTASs generally follow a cultural dissemination model that conforms to mountains, waters, and other natural barriers. For instance, the northern Jing and Jin styles cluster in the Yellow River Basin, the southern Su and Wan styles follow intricate water networks, and the Min and Chuan styles are dispersed across hilly and mountainous terrain. At the mesoscale, Local Indicators of Spatial Association (LISA) analysis identified statistically significant high-high clustering areas for each style. Synthesizing these with topographic and hydrological data allowed for the delineation of coherent cultural geography: five distinct cultural core areas (where a single style is intensely concentrated), three well-defined cultural transition belts (exhibiting high stylistic mixing), and two cultural fracture zones (areas of low style density and diversity). This regionalization provides a precise spatial template for designing cross-provincial heritage corridor protection systems. At the micro (provincial) scale, the diversity and evenness indices revealed three characteristic patterns: Multi-Style Fusion Zones (high diversity, high uniformity), Oligarchic Equilibrium Zones (low diversity, high uniformity), and Style-Pure Zones (low diversity, low uniformity). These patterns offer crucial insights for tailoring provincial-level conservation strategies and regional planning. This study makes significant strides on these two fronts. Methodologically, it pioneers a replicable and scalable framework that successfully bridges computer vision and cultural geography, offering a new and robust tool for automated architectural stylistic analysis. Substantively, it provides the first comprehensive, quantitative, and multiscale spatial dissection of CTASs based on large-scale empirical data. Collectively, these findings provide a rigorous scientific foundation for evidence-based policies in multiscale heritage conservation, cultural landscape management, and sustainable regional development.
Traditional geography teaching has long faced core challenges, such as difficulties in visualizing abstract concepts and the limitations of high-risk or high-cost practical activities. Although AI-generated content (AIGC) and metaverse technologies have individually shown educational potential, existing research is largely confined to single-technology applications. There is a lack of a cross-technology synergistic framework tailored to the three-dimensional attributes of geography ("spatial-temporal-process"), resulting in insufficient adaptation between technology and pedagogical needs. To address this gap, this study aimed to construct a novel geography teaching model synergistically empowered by "AIGC + Metaverse". Through empirical research, it sought to validate the effectiveness of the proposed model in enhancing students' core competencies, thereby providing a verified and systematic solution to overcome traditional teaching dilemmas an."d promote the intelligent transformation of geography education. This study developed a closed-loop teaching model consisting of "Scenario Immersion—Intelligent Guidance—Assessment Optimization". The model employed metaverse technology to create high-fidelity immersive geography scenarios (e.g., a digital twin environment of a dangerous rock mass), while leveraging AIGC to build an intelligent tutoring system that supported personalized guidance, intelligent Q&A, and formative assessment. Deep integration was achieved through data and pedagogical process synergy mechanisms. To validate the model's effectiveness, a quasi-experimental study was conducted over two academic years using the "Spatial Information Collection and Risk Assessment of Dangerous Rock Mass" experiment as a case study. A multidimensional dataset, including experimental scores, learning behavior data, and system logs, was collected to comprehensively compare outcomes between a traditional teaching group and an "AIGC + Metaverse" synergistic group. The empirical results demonstrated that the "AIGC + Metaverse" synergistic teaching model achieved significant improvements over the traditional model. In terms of practical operation ability, students' average scores in operational standardization and data completeness both increased by 12.5%. Improvements in spatial analysis ability were particularly prominent, with the average score for measurement accuracy increasing by 28.57% and the average score for analytical logic by 14.29%. Regarding innovative application ability, the average score for the quality of risk assessment reports rose by 28.57%, and the average score for thinking questions by 14.29%. Ultimately, the average score for comprehensive task ability improved by 11.25%. The study concluded that this synergistic model, by combining immersive metaverse scenarios with the intelligent AIGC feedback, effectively resolved the core contradictions of traditional geography teaching. It facilitated a shift in students' roles from passive recipients of knowledge to active explorers of capability, thereby achieving a fundamental paradigm shift from "knowledge transmission" to "capability empowerment." The contributions of this study are threefold. First, it offers theoretical innovation by constructing and validating a synergistic teaching framework integrating AIGC-generated content, metaverse-hosted scenarios, and real-time AI tutor interaction, thus filling the gap in existing research on cross-technology synergy and discipline-specific adaptation. Second, it provides a practical paradigm by delivering a replicable, scalable, and systematic teaching solution, serving as a concrete example for geography education reform. Third, it contributes to domain expansion by exploring a viable path for the deep integration of advanced intelligent technologies with subject teaching in the context of educational digitalization, offering theoretical references and practical insights for innovative teaching in related disciplines.
