Nature journal publishes Harbin Institute of Technology study on generative AI's 'triple identity' in public governance

A study published on 12 October 2026 in the journal Humanities and Social Sciences Communications examines how generative AI affects public governance. Its authors, Sihan Lu, Yan Zhou and Tao Hong of the Harbin Institute of Technology, say existing research tends to treat generative AI in isolation, as a governance tool, as something to be regulated, or as a new kind of decision-maker, which hides the paradoxes and systemic risks that appear when these roles overlap.

To address that gap, the authors model public management as a complex adaptive system and treat generative AI as an adaptive agent within it. They use a single case study, Singapore's Smart Nation initiative, and track governance strategies and results across different situations to draw general lessons about the technology's triple identity.

The findings describe a Governance Impossible Triangle, a lasting tension among maximizing efficiency, protecting the legitimacy of decisions and controlling risk. As a way through, the paper proposes a Contextual Identity Configuration Model shaped by how complex the problem is, how severe the risk is, clashing values and explainability demands. It applies the model to four typical settings, which range from routine optimization to complex crises and include professional assistance and high-stakes discretion. In each, management systems adjust how far AI works as a tool versus an agent with power, through continuing feedback loops.

The authors conclude that policymakers should abandon one-size-fits-all regulation for an agile, context-based approach, using layered defenses and shared accountability networks so that technological empowerment stays aligned with public values. Funding came from a Chinese national science foundation and a Heilongjiang provincial office.