摘要
当前,生成式人工智能(GAI)正驱动犯罪侦查模式向预见性情报主导范式跃迁。以风险传导的动态博弈体系为理论内核,依托生成对抗网络、时空图神经网络与多智能体强化学习构建的“风险识别-传导模拟-策略优化”闭环模型,可揭示犯罪能量在复杂系统中的非线性传导规律,进而形成数据联邦共享、数字孪生模拟、混合博弈策略与智能联动预警的闭环优化体系,解决数据孤岛、模型僵化、博弈失衡与预警缺陷等矛盾。由此,不仅可以推动犯罪治理从被动追溯转向主动干预范式,也将为社会治理能力现代化提供兼具算法正义与治理效能的理论范式,对全球犯罪治理新秩序的形成具有战略价值。
Currently,generative artificial intelligence(GAI)is driving the transformation of criminal investigation models towards a predictive intelligence-led paradigm.Based on a dynamic game system of risk transmission as its theoretical core,a closed-loop model constructed with generative adversarial networks,spatio-temporal graph neural networks,and multi-agent reinforcement learning for“risk identification-transmission simulation-strategy optimization”can reveal the nonlinear transmission laws of criminal energy in complex systems.This,in turn,forms a closed-loop optimization system of data federation sharing,digital twin simulation,hybrid game strategies,and intelligent linkage early warning,resolving issues such as data silos,model rigidity,game imbalance,and early warning deficiencies.Consequently,it not only promotes the shift of crime governance from passive tracing to proactive intervention but also provides a theoretical framework that combines algorithmic justice and governance efficiency for the modernization of social governance capabilities,which holds strategic value for the formation of a new global crime governance order.
作者
赵长明
刘欢
ZHAO Chang-ming;LIU Huan
出处
《江苏警官学院学报》
2025年第3期67-77,共11页
Journal of Jiangsu Police Institute
基金
公安部科技计划理论软科学项目(2022LL43)
陕西省“十四五”教育科学规划课题“双高计划引领下多元共建公安高职院校教学创新团队建设研究”(SGH23Y3093)
陕西省高等教育教学改革研究项目“高职侦查类一流核心课程建设与应用研究”(23GY052)
西安市科技计划理论软科学项目“数字赋能西安公安机关预防化解城市社区矛盾纠纷机制建设研究”(25RKYJ0078)。
关键词
GAI
犯罪情报
主动式侦查
GAI
crime intelligence
proactive investigation
