Join us for an insightful HiveEngage session led by Phil Baker, who will draw on CADi’s work in causal governance to explore one of the most significant risks facing organisations as they adopt AI.
Confidently Wrong: The AI Failure Mode Your Governance Can’t See
Wednesday, 7 October
Time: 11:00 – 12:00
Live via Google Meet
Why Attend?
Most AI governance can tell us what a system decided and how it reached that decision. However, it often cannot answer the question that matters most when something goes wrong:
Why did this happen and what would have happened if the AI had decided differently?
This gap is where real harm can occur. Decisions that are confidently wrong may remain invisible to dashboards and explainability tools, disproportionately affecting vulnerable customers who are least equipped to challenge them.
In this practical session, Phil will introduce causal governance and explain how causal inference and counterfactual analysis can make AI decisions more auditable, contestable and capable of meaningful improvement.
Drawing on CADi’s work over the past year including testing through the FCA’s Digital Sandbox, Phil will share examples from payments, agentic AI workflows and the public sector. These demonstrate how an AI model can appear accurate while still causing harm to customers.
This will be a practical discussion, not a maths lecture, and no technical background is required.
What Will You Learn?
- The counterfactual gap: Why monitoring and explainability can show what happened, but not necessarily what should have happened.
- The five questions test: A practical framework you can use to assess any AI deployment:
- What happened?
- Why did it happen?
- What would have happened otherwise?
- What should happen next?
- Can you prove it?
- Where regulation is heading: How Consumer Duty, the FCA’s work on AI and the EU AI Act are increasingly focusing on customer outcomes and evidence of protection for vulnerable customers.
Who Should Attend?
This session will be particularly valuable for:
- Consultants and programme directors advising clients on AI adoption
- Product and technology leaders
- Chief Risk Officers and risk and audit professionals
- Data scientists
- Change and transformation specialists
- Professionals working in regulated or high-stakes sectors, including financial services, government, defence and healthcare
Date/Time
Date(s) - 07/10/2026
11:00 am - 12:00 pm
Timezone: Europe/London
Bookings
Categories
