The talk in brief
Fluent systems fail more persuasively. When AI sounds confident, customers believe it, and so do the people governing it. The gap between fluency and competence is where trust is won or lost.
This keynote examines what responsible AI in customer experience actually requires, from conversation design to governance to the human experience of interacting with a system that sounds right but may not be. Its central argument: responsible AI is not a governance checkbox. It is a design practice that starts before the model is chosen.
What it covers
Perceived comprehension
How AI creates the appearance of understanding while failing the customer, and why that is harder to detect than an obvious error.
The moment of failure
The most important moment in any AI interaction is not when it works. It is when it fails and the handoff to a human has to happen. Most escalation paths are designed last. They should be designed first.
Who is accountable
Who owns the AI, who governs it, and who answers for it when it behaves unexpectedly. Most organisations don't yet have a clear answer.
Metrics that work against the customer
How containment rate, CSAT, and handle time can create systems that optimise against the people they are meant to serve.
What the audience leaves with
- A practical definition of responsible AI in CX that design, risk, and operations teams can share
- A way to spot the subtle failures that fluent AI hides
- Principles for escalation and human-in-the-loop design that protect trust
- Questions for governance forums that go beyond policy documents
This keynote draws directly on Designing AI Conversations at Scale.
Frequently asked questions
What is Ben Farrell's responsible AI keynote about?
It examines what responsible AI in customer experience requires, from conversation design and escalation to governance and accountability, and argues that responsible AI is a design practice that starts before the model is chosen.
Is the responsible AI keynote based on Ben Farrell's book?
Yes. It draws directly on Designing AI Conversations at Scale, and a workshop extension is available for design, risk, and leadership teams.