What does it take for an AI-powered digital twin to support a decision that people can trust? In AI-Powered Common Data Models & Governance for Trustworthy Digital Twins – Building Resilient Ecosystems, the panel looks beyond access to data and examines whether that data has the meaning, consistency and oversight needed for industrial use.
The discussion begins with a practical challenge: organisations cannot wait until every record is perfect before using AI. Starting with a defined operational problem helps identify which data is essential and where improvements will make the greatest difference. The speakers also stress that errors found during everyday work should be corrected at their source where possible. Bringing more information together is useful only if conflicting records and the relationships between them can be understood.
Common data models, taxonomies and semantic context can help connect information across assets, teams and suppliers. Yet the panel makes clear that choosing a model is itself a governance decision. Data owners and operational teams need to agree on what information means, who can access it and how it should be used within a workflow. Those decisions become more consequential when AI moves from finding information to recommending or taking action.
The speakers consider where human judgement, business rules and technical guardrails belong in that process. AI can help coordinate information across systems, but a plausible answer is not necessarily a safe one. For decisions with significant physical or safety consequences, the panel explores the value of clear access permissions, defined operating limits and checks against more constrained models. They also discuss the risk of human reviewers approving recommendations routinely without assessing them.
Finally, the conversation turns to workforce trust. People who work with industrial data often know its gaps and informal workarounds, knowledge that can disappear when information is combined across systems. Making data sources and limitations visible, inviting users to challenge results and maintaining skills as routine tasks are automated all help build confidence over time.
Watch the full panel discussion to hear the speakers’ perspectives and explore the examples and questions behind these themes.
Speaker
- Mesbah Khan, Data Innovation Manager, Shell
- Ajish Varghese, Technology Lead, Harbour Energy
- Tomasz Wiktorski, Principal Program Engineer, Future Data Operations, Aker BP
- Shane McArdle, CEO, Falkor
- Talo Dourado Affonso, Digital Asset Maintenance, Petrobras
Moderator
- Jorge J. Buendia, Global Digital Transformation & AI Lead, BASF
