AI in Governance
Authors: Jin Xue, Wei-Ting Hong, Qian-Cheng Wang, Yunping Liang, Jennifer Whyte and Kerui Lyu
This research develops an AI-driven tool to support stakeholder management in large infrastructure projects. It uses large language models and knowledge graphs to extract and structure information from project documents. This includes identifying who was involved, what actions were taken, and how engagement changes over time. The tool compares planned engagement with actual outcomes, which helps project teams understand stakeholder issues more clearly and supports more informed decision-making.
Why is this research relevant?
Large infrastructure projects involve multiple stakeholders at different stages. Their interests and influence change over time, especially on ESG issues. Traditional methods often fail to capture these changes, and this can lead to differences between planned and actual engagement. This model uses AI to improve this, it makes stakeholder activity easier to see and understand, while keeping track of changes over time. This supports better monitoring, clearer insights and improved decisions in complex projects.
Who should read this report?
- Project managers working on mega-infrastructure projects dealing with complex stakeholder environments
- Programme and portfolio directors responsible for strategy and delivery oversight
- ESG and sustainability professionals focusing on environmental, social, and governance reporting and compliance
- Public sector decision-makers involved in infrastructure planning, approval and accountability
- Academic researchers in project management, AI, and data-driven decision support systems
- Consultants and advisory firms supporting infrastructure delivery and stakeholder engagement strategy
How was the research undertaken?
The research was carried out using two types of methods: case studies and data analysis. It looked at official project documents from Australia, the UK, and the US. AI tools called large language models were used to find information about stakeholders and key issues, while knowledge graphs were used to show how stakeholders and events are connected. Researchers also gave scores to show how much engagement was needed for ESG issues. These were compared with actual engagement found by AI. The results were shown in a visual map and built into a web platform.
What did we discover?
The analysis showed that stakeholder engagement in mega-projects is often uneven, with clear cases of alignment, over-engagement, and under-engagement across ESG-related issues. Knowledge graphs revealed how stakeholders connect to specific events and how these relationships evolve over project phases. The strategic engagement maps highlighted gaps between intended and actual engagement, showing where resources were misallocated or insufficient. Overall, the results demonstrate that AI-driven analysis can uncover hidden patterns in stakeholder dynamics that are not easily visible in traditional reporting methods.
Key recommendations?
- The findings suggest that project teams should pay closer attention to gaps between planned and actual stakeholder engagement. This helps identify over-engaged and under-engaged areas across ESG issues. Attention should be shifted where engagement does not match project needs.
- The study also supports using knowledge graphs to track how stakeholder relationships change across project phases. This helps improve visibility of complex dynamics over time.
- AI tools can also be used to surface evidence directly from project documents. This improves transparency and supports more informed, traceable decision-making in stakeholder management.