Ninety-seven percent of Emiratis use AI at work, in study, or in daily life. Fifty-three percent believe its benefits outweigh its risks, a confidence level that puts the UAE ahead of much of the world on AI acceptance. And yet, when a UAE resident picks up the phone or opens a chat window to resolve a genuine customer service problem, 87% say they would rather speak to a real person than a chatbot or AI system.
Those two facts come from different studies. The first from KPMG’s 2025 global “Trust, Attitudes and Use of AI” research covering more than 48,000 people across 47 countries; the second is from a YouGov-fielded survey of 1,000 UAE residents commissioned for Zbooni’s 2024 MENA cCommerce Report. Read side by side, they describe something sharper than mild skepticism. This is a region that has embraced AI more enthusiastically than most of the world, while simultaneously drawing one of the clearest lines anywhere against letting that same AI handle a service conversation that matters. Call it the trust paradox. Understanding why it exists and exactly where it does and doesn’t apply is now a more urgent question for GCC customer service leaders than any deployment roadmap.
Where the Paradox Resolves and Where It Doesn’t
Before unpacking why this gap exists, it’s worth resolving the apparent contradiction, because the answer changes what a business should actually do about it.
The paradox dissolves once you stop treating “AI in customer service” as one thing. McKinsey’s 2025 research on AI in the GCC found that adoption has surged from 62% of organizations in 2023 to 84% in 2025, nearly matching the 88% global average, but only 31% of those organizations have moved beyond pilots to real scale, and just 11% qualify as genuine “value realizers” capturing measurable earnings impact from AI. That gap between adoption and value maps almost exactly onto the gap between AI usage and AI trust in customer service. People and organizations alike are willing to use AI broadly and are far more cautious about depending on it for anything that carries weight.
Applied to customer service, this produces a usable segmentation rather than a blanket verdict. Routine, low-stakes, transactional interactions such as checking a balance, confirming an appointment, tracking an order, are exactly where GCC adoption and trust already converge; these are precisely the use cases we identified as where AI voice agents “earn their keep” in our earlier look at the UAE market. Complex, emotional, financially significant, or ambiguous interactions are where the 87% preference for a human holds in full force, largely undiminished by how sophisticated the AI has become. The paradox, in other words, isn’t a verdict on AI’s readiness. It’s a map of exactly where GCC businesses should be automating first and where they still need a person on the line—a map most vendors flatten into a single “AI adoption” number that obscures the very distinction that determines whether a deployment succeeds.
Why This Isn’t a Contradiction. It’s a Signal
Once you see adoption and trust as two separate variables, the paradox stops looking irrational and starts looking like useful information. Verloop’s research on UAE customer support automation found that 76% of support teams invested in AI in 2024, up from 54% the year before, with 83% planning to increase automation spend further, yet 71% of consumers in the same market said they still preferred a human for complex or emotional issues. Businesses are moving faster than their customers’ comfort, and the direction of that gap matters: it is not that GCC customers reject AI outright, but that they are more discerning than global averages about where they’ll extend their confidence.
This discernment shows up again in a widely cited finding from Retell AI’s research on call abandonment: only 7% of customers say they trust AI most for resolving an issue, compared to 51% who trust a human agent most, a gap that dwarfs the productivity case most vendors lead with. And it isn’t just a regional quirk. Global research on AI customer service cited by Gladly found that while 88% of AI-handled issues are reported as “resolved,” only 22% of customers said the experience made them prefer the company afterward. A 66-point gap between deflection and loyalty that shows this dynamic playing out well beyond the GCC. What’s distinctive about the GCC is not that this gap exists but how much more sharply it’s expressed here: 87% choosing a human is a considerably firmer number than most Western CX benchmarks report for the same question.
What “Trust” Actually Breaks Down Into Here
It would be easy to read all this as a region uneasy about AI generally. The data says the opposite. KPMG’s UAE-specific findings show 68% of respondents believe current AI regulation is sufficient to ensure safe use as opposed to the 43% global average, even as 88% want stronger laws specifically against AI-generated misinformation. Put together, these numbers describe a public that trusts the system being built around AI considerably more than it trusts AI in the moment of a personal, one-to-one interaction. That is a meaningfully different problem than “customers don’t trust AI,” and it points to a different fix: the issue isn’t regulatory confidence or general acceptance, both of which are already high. It’s what happens the moment a customer’s own money, health, or unresolved frustration is on the other end of the conversation.
The Channel Clue
A second, easily missed data point sharpens this further. The same Zbooni/YouGov research found that 65% of UAE residents had contacted a business via WhatsApp in the past year, more than any other channel, ahead of call centers at 55%, email at 48%, and social media at under 30%, with 85% wanting every business to offer WhatsApp support and 88% believing it delivers the quickest, most accurate responses. Read next to the 87% preference for human interaction, this isn’t a contradiction, either. WhatsApp isn’t popular here because it feels automated; it’s popular because it feels personally attended to: asynchronous, direct, and unmistakably a conversation rather than a queue. As Zbooni’s CEO Ramy Assaf put it, describing the same data: “People still prefer genuine human interaction over bots or AI. While there’s a place for automation, especially for handling simple, repetitive tasks, humans still respond with more empathy, warmth, and personalization.” The lesson for AI deployment isn’t to avoid the channels customers already prefer; it’s to make sure whatever operates in them, human or AI, preserves that same sense of being personally attended to.
Evidence From Brands Already Closing the Gap
This gap is not fixed, and KPMG’s UAE customer experience research points to who’s closing it. Across 104 UAE brands tracked, KPMG found that integrity has overtaken personalization as the strongest driver of customer experience in the country for the first time with trust, transparency, and empathy explicitly named as the qualities “taking centre stage in the AI era.” The brands rising fastest in that ranking are instructive: Emirates Islamic Bank climbed 42 places, and Careem 32, alongside long-standing leaders like Emirates and Etihad. None of these are companies winning by automating the most. They’re winning, per KPMG’s own framing, by making automation feel like it still has a person’s judgment behind it, which is precisely the distinction the 87% figure is measuring.
What This Means for a GCC Business Deploying AI Voice or Chat
Three principles follow directly from the data above, and they apply whether the channel is voice, chat, or WhatsApp. First, sequence deployment by where trust and adoption already overlap, such as the routine, resolution-oriented interactions identified in the segmentation above, rather than pushing AI into complex or emotional conversations before it has earned standing there. Second, disclose deliberately rather than incidentally: given how sharply UAE consumers separate their confidence in AI-as-a-system from their comfort with AI-in-a-conversation, telling a customer plainly that they’re speaking with an AI and why that’s appropriate for this interaction does more to build trust than trying to make the AI convincingly human. Third, design for attentiveness, not just speed. The WhatsApp data makes clear that GCC customers reward channels and interactions that feel personally considered, and an AI system optimized purely for resolution time, without that quality will keep running into the same 87% wall.
The Real Takeaway
The trust paradox is not a reason for GCC businesses to slow down AI adoption. The region is already moving faster on that front than most of the world, and the underlying comfort with AI as a technology is genuinely high. It is, instead, the clearest evidence yet that customer service AI in this market needs to be more deliberately sequenced, more transparently disclosed, and more carefully designed for warmth than the playbooks written for other regions assume. The businesses that read this data correctly won’t wait for the paradox to resolve itself. They’ll use it as the most precise map available for exactly where to build next.
Frequently Asked Questions
1.Does this data mean GCC businesses shouldn’t invest in AI voice or chat?
No. It means investment should be sequenced deliberately. The same research shows strong AI usage and adoption; the caution is specific to complex or emotionally significant service conversations, not to AI broadly.
2.Is this a UAE-only pattern or true across the GCC?
The hardest numbers cited here (KPMG, Zbooni/YouGov) are UAE-specific, but the underlying dynamic of adoption outpacing trust shows up in Verloop’s UAE support-automation data and in McKinsey’s broader GCC-wide finding that only 11% of organizations are realizing measurable AI value despite 84% adoption. The pattern is regional even where the precise figures are national.
3.How do we know if our own customers are in the “trust gap” zone?
Look at where your escalation and complaint volume actually concentrates. If dissatisfaction clusters in complex, financial, or emotionally charged interactions rather than routine ones, that’s a strong sign you’re deploying AI ahead of where trust currently sits for that use case. Its a segmentation problem, not an AI-quality problem.
4.Doesn’t disclosing that a system is AI reduce customers’ willingness to use it?
The data here suggests the opposite risk is larger: KPMG’s finding that UAE consumers trust the regulatory system around AI more than they trust it in personal interactions suggests disclosure builds on existing institutional confidence rather than undermining it, particularly when paired with a clear, easy path to a human.
