For decades, the customer experience industry’s central challenge has been fixing bad service. Rude agents, long wait times, unresolved complaints.
These problems were well understood, and organisations knew broadly how to address them. The damage they caused was real but slow, and recovery was possible.
CEM Africa.com | View the event programme | Ticket options
Bad AI is different. It moves faster, scales instantly, and erodes trust in ways that are harder to detect and far harder to repair.
And in 2026, as organisations across Africa and the world accelerate AI deployment in customer-facing roles, the question is not whether AI will be used. The question is whether it will be governed well enough to protect the trust it is supposed to build.
This is the provocation at the heart of CEM Africa 2026’s session on the trust gap: exploring why poorly designed, hastily deployed and inadequately governed AI may represent the single greatest threat to customer trust in the decade ahead.
The statistics paint a picture that should concern every CX leader. Although 66% of people are already using AI with some regularity, fewer than half are willing to trust it.
Only 15% of adults trust companies that use AI in customer interactions. And when things go wrong, the cost is steep: when an AI interaction fails to resolve a customer’s issue, the Net Promoter Score for that interaction can plunge by as much as 70 points.
Consumer anxiety is not abstract. According to research by Relyance AI, 82% of consumers see AI data loss-of-control as a serious personal threat.
Over 80% believe AI-generated content should be clearly labelled, and 62% say that transparency would increase their trust in a brand. Yet most organisations are deploying AI in customer journeys without either the labelling or the transparency customers are asking for.
The gap between what organisations are building and what customers are prepared to trust is widening at exactly the moment when AI deployment is accelerating.
According to Sinch research, 62% of organisations already have AI agents live in production across customer channels, with 88% expected to follow by the end of 2026. The infrastructure is moving far ahead of the governance.
Customers are not afraid of AI. They are afraid of AI they cannot understand, cannot challenge, and cannot escape when it goes wrong.
ALSO READ: Why your customer experience is your biggest growth risk
Traditional service failures are usually visible. An agent handles a complaint poorly. A queue is too long. A product is out of stock. These failures are traceable, measurable, and correctable through familiar levers: coaching, resourcing, process redesign.
AI failures are structurally different. A 2025 MIT study found that AI models are 34% more likely to use high-confidence language, words like “definitely” or “certainly,” when delivering incorrect information than when delivering accurate information.
In other words, AI is most convincing precisely when it is most wrong. This creates a specific and serious risk in customer-facing contexts, where customers may act on confidently stated misinformation before any error is detected.
The chatbot failure rate reinforces this. Research from Sinch found that 74% of AI customer service chatbots are pulled offline or rolled back after launch due to failures.
The most common causes are not technical breakdowns but design failures: systems that cannot recognise the boundaries of their competence, cannot escalate gracefully, and cannot recover from edge cases that fall outside their training.
The result is customer interactions that feel worse than no service at all.
The McDonald’s AI drive-through pilot, which was ultimately shut down after viral videos captured a system adding 260 Chicken McNuggets to a single order, became a widely cited example.
But the more common AI failures are less dramatic and therefore more dangerous: small misrepresentations, failed escalations, responses that are technically plausible but contextually wrong.
These failures compound over time, eroding trust gradually until customers simply stop engaging.
Until recently, the dominant narrative in AI and CX has been about speed and efficiency. How quickly can AI handle volume? How many interactions can it deflect? What percentage of tickets can it resolve without a human?
By 2026, that conversation is shifting. According to McKinsey’s State of AI Trust report, in 2024 and 2025, organisations focused on increasing automation.
The 2026 focus is shifting to accountability. Only 6% of companies fully trust AI agents to handle core business processes.
Most organisations restrict AI to routine or supervised tasks, which suggests a growing awareness that deployment speed has outpaced governance maturity.
KPMG’s global AI trust research confirms the challenge: only about one-third of organisations report any maturity in AI governance and oversight structures.
Technical capabilities are advancing faster than the organisational systems designed to supervise them. And in customer-facing contexts, this gap is not an internal risk, it is a customer trust risk.
The organisations that will win on customer trust are not those with the most AI. They are those with the most accountable AI.
Research consistently shows that customers do not demand perfection from AI. What they demand is honesty and recovery.
COPC’s global consumer research found that satisfaction with AI interactions rises above 90% when the interaction resolves the issue without further steps.
The problem is not that AI exists in the journey; it is that AI sometimes pretends to be capable of things it is not, or fails to offer a clear path forward when it cannot help.
Transparency addresses this directly. Over 80% of consumers want AI-generated content labelled. Customers are more willing to accept a “no” when they understand why.
And according to research by RWS, the first organisation in an industry to demonstrably prove its AI transparency captures 76% of the addressable market willing to switch.
In other words, governance and openness are not just ethical obligations, they are competitive advantages.
CX leaders are beginning to recognise this. 87% of CX leaders now agree that AI transparency will be non-negotiable for customer-facing AI within two years.
72% of CX leaders in the UK say it is mission-critical that AI systems can show their reasoning. The question is no longer whether to be transparent about AI, it is how fast organisations can build the governance frameworks that make transparency possible.
Accountable AI in customer experience is not a single feature or a compliance checkbox.
It is a set of design principles applied consistently across the AI lifecycle: from how models are selected and tested, to how they are monitored in production, to how failures are detected and escalated.
Practically, it means AI systems that know their own limits and say so. Escalation paths that are clearly labelled and easy to reach. Feedback mechanisms that surface failure modes before customers stop complaining and simply leave.
Governance frameworks that sit inside CX operations, not just in a separate technology team. And leadership accountability that treats AI failures in customer journeys with the same seriousness as agent failures.
It also means rethinking how AI success is measured. Deflection rates and resolution percentages matter, but they do not capture the trust impact of a failed interaction. Organisations that measure only what AI resolves are missing what AI breaks.
The cost of a 70-point NPS drop in a single interaction is not visible in a deflection dashboard.
The African context adds specificity to this challenge. Across diverse markets with varying digital literacy, data connectivity, and consumer protection frameworks, the risks of poorly governed AI are not evenly distributed.
Customers in lower-data environments or with limited digital confidence are least able to challenge AI responses and most likely to be harmed by confident misinformation. Designing for trust is not just a strategic question, it is an equity question.
Sessions like this one matter because they place governance, transparency and accountability at the centre of the AI conversation, not as a compliance footnote but as the foundation of customer trust.
CEM Africa 2026 brings together the practitioners responsible for building and managing AI-enabled customer experiences across the continent, to wrestle with the real questions: how do you govern AI well when deployment pressure is high?
How do you balance speed with accountability? And how do you ensure that AI makes customer trust stronger, not more fragile?
These are not questions with easy answers. But they are questions that need to be asked out loud, by the right people, in the same room.
READ NEXT: Why call centre CX belongs at the centre of CEM Africa 2026
Contact Us
VUKA is the trusted media partner to key professionals, policy makers, suppliers and
manufacturers. We provide unparalleled opportunities for industry-wide connection.
