Abstract
Amid the unprecedented wave of AI advancement, AI-resistant professional skills play a significant role in enhancing the effectiveness of human-AI collaboration. However, existing research tends to isolate professional skills from their broader context, overlooking the triadic construction of digital identity recognition through individual motivation, structural position, and knowledge articulation. This oversight weakens the sustainability and adaptability of skill expression, thereby hindering innovation performance in AI-HI (Artificial Intelligence-Human Intelligence) collaboration. Drawing on the entropy weight method, gradient descent algorithm, and a residual-matching decision matrix, this study conducted quantitative modeling of 418 participants in the financial co-production sector from 2022 to 2024. The findings reveal that network centrality (NC; β = 0.04(**)) and proactive personality (PP; β = 0.05(**)) significantly amplify the impact of two key AI-resistant skills-foreign language proficiency (FL) and passion/optimism (PO)-on collaboration effectiveness, through structural empowerment and intrinsic motivation. Furthermore, this study develops a digital identity recognition and classification framework that identifies three distinct groups: core innovators, marginal experts, and low performers. By extending the theoretical model of digital identity construction within AI-HI collaboration, this study also proposes a differentiated approach to talent development and resource allocation based on innovation effectiveness and identity alignment, offering new insights into the advancement of digital human capital.