地方政府推动人工智能发展的专项政策质量评估——基于“目标-路径-工具”组合框架的PMC分析

Quality assessment of local governments" special policies to promote AI development-- PMC analysis based on "goal-path-tool" combination framework

摘要:

省级人工智能专项政策是引领地区人工智能良性发展赋能经济社会转型升级的关键驱动力。厘清省级人工智能专项政策的内在逻辑与质量水平，对各地区优化人工智能政策举措具有重要意义。基于“目标-路径-工具”组合框架，利用PMC指数模型对中国24份省级人工智能专项政策进行文本量化分析。研究发现：（1）政策目标上，现行省级人工智能专项政策包括时序维度的近期、中期、远期3类目标设定和基础理论、应用技术、企业发展、产业样态、规模预期、人才队伍、政策法规7种面向发展向度的目标预设；（2）政策路径上，省级人工智能专项政策存在创新设施、科技研发、产业发展、人才队伍、发展环境和资助体系六大政策路径；（3）政策工具上，目前省级人工智能专项政策的工具设置，遵循“产出工具＞资源工具＞保障工具＞导向工具”的价值侧重；（4）质量评价上，省级人工智能专项政策质量水平横跨优秀、良好、合格和不良4个层级，而平均水平仅处于合格水平，同时，政策质量评估指标体系的7个一级指标得分存在“科技研发＞产业发展＞创新设施＞发展环境＞资助体系＞人才队伍＞政策目标”的情况。据此提出省级地方政府需要因地制宜地优化人工智能专项政策，依据政策目标、路径和工具要素方面存在的不足，推动政策各要素均衡配置，从而提升政策质量水平与实施效能。

Abstract:

Provincial-level special policies on artificial intelligence (AI) serve as a key driving force in guiding the healthy development of regional AI and empowering economic and social transformation and upgrading. Clarifying the internal logic and quality levels of these policies is of great significance for optimizing AI policy measures across different regions.Based on the goal-path-tool framework, this study employs the PMC index model to conduct a quantitative textual analysis of 24 provincial-level AI special policies in China. The findings are as follows:(1) In policy goals, the current provincial-level AI special policies establish three types of temporal objectives:short-term, medium-term,and long-termalongside seven developmental dimensions, including fundamental theory, applied technology, enterprise development, industrial patterns, scale expectations, talent cultivation, and policy and regulations.(2) In policy pathways,these policies follow six major pathways: innovation infrastructure, scientific and technological research and development (R&D), industrial development, talent cultivation, development environment, and funding systems.(3) In policy instruments,the design of policy instruments follows a prioritization hierarchy of "output tools > resource tools > safeguard tools > guiding tools".(4) In policy quality assessment, the quality levels of provincial AI special policies vary across four categories:excellent, good, qualified, and substandard,yet the average quality level only meets the qualified standard. Additionally, the scoring of the seven primary indicators in the policy quality evaluation system follows the order of "scientific and technological research and development > industrial development > innovation infrastructure > development environment > funding systems > talent cultivation > policy goals". Accordingly, this study suggests that provincial governments should optimize AI special policies based on local conditions, addressing deficiencies in policy goals, pathways, and instruments. By ensuring a balanced configuration of these elements, governments can enhance policy quality and implementation effectiveness.