人机协同视域下吐鲁番景观价值评价与智慧服务响应机制OA

[Objective]This study addresses key challenges in the evaluation of arid and semi-arid landscapes in western China,such as vast spatial extent,high fieldwork costs,insufficient sample coverage,and delayed service responses.It aims to construct a novel low-cost"human-AI coupling"paradigm for smart service generation in order to tackle the spatiotemporal constraints of conventional public participation and provide a basis for the target-centered governance of the underdeveloped west region.[Method]We develop a dual-track model of"AI prediction+ground-truth validation".Large language models(LLMs)are used to generate virtual public agents whose demographic structure is isomorphic to that of the surveyed population,thereby establishing an AI-based value baseline.Field questionnaires are collected as ground truth to enable coupling-based comparison,discrepancy quantification,and diagnostic analysis.[Result]Significant divergences(q<0.05)are identified between AI predictions and in situ public ratings.The AI tends to overestimate knowledge-oriented values(e.g.,scientific and historical values),whereas public evaluations are more strongly shaped by factors such as on-site experience and differentiation in visitors/residents'identity and educational attainment.Discrepancy diagnostics further reveal three spatial-experiential patterns,i.e.,"on-site immersion-for-gain zone","cognitive deficit zone"of high-value but low-perception,and"service blind spot"of high-expectation but low-experience.[Conclusion]Based on the diagnostic biases,we propose smart response mechanisms including AR-enabled heritage interpretation,microclimate-sensing dynamic touring routes,and a community digital memory platform.The findings verify that AI can serve as a rational reference and an effective tool for detecting service blind spots in the early stages of planning.Furthermore,future research should introduce large multimodal models(LMMs)in connection with microclimate data to establish a"human-in-the-loop"reverse calibration mechanism,driving the transition from static evaluation to dynamic and adaptive governance.