Fifty-six percent of people say they trust an AI chatbot to act in their best interest. Only 3.7% would let that same chatbot make a decision for them. If those two numbers don't make you pause mid-sip of your morning coffee, you're not paying attention.
That gap, surfaced in recent research from the Collective Intelligence Project's Global Dialogues, tells us something uncomfortable: trust in AI is arriving faster than our ability to verify whether that trust is deserved. And for marketers, this creates a paradox that no martech stack can solve.
We've spent the last two years watching AI become the ultimate productivity multiplier. Generate emails in seconds. Personalize landing pages at scale. Spin up ten campaign variations before lunch. The capability is intoxicating. But here's the thing about capability: it doesn't care whether your organization can actually deliver on what it promises.
The Velocity Problem
Julie Zwissler, writing for MarTech, put it perfectly: "AI can help you make promises faster than your organization can build the trust needed to deliver on them." That sentence should be tattooed on every CMO's forearm.
Think about what AI enables in a typical marketing operation. You can now reach more customers, with more personalized messages, across more channels, in less time than ever before. Sounds like a dream, right? Except every one of those touchpoints is a promise. A promise that you understand the customer. A promise that you'll deliver value. A promise that when something goes wrong, a human being will actually show up.
And promises, unlike content, don't scale.
The data backs this up. According to a Gartner survey of 1,464 B2B buyers and consumers, 53% experience negative outcomes from traditional personalization, and those customers are 3.2 times more likely to regret their purchase. Let that sink in: personalization done poorly is now measurably worse than no personalization at all.
The Trust Penalty Is Real
Here's where it gets expensive. Research from Klaviyo and Datalily found that only 7% of consumers say visible AI-generated marketing content makes them trust a brand more. Meanwhile, 31% say it makes them trust the brand less. That's not a gap; that's a canyon.
And consumers aren't just feeling skeptical. They're acting on it. A survey of 11,000 consumers
found that 47% took at least one action with a direct revenue consequence because of how a brand handled their data and AI. One in four canceled a subscription or stopped purchasing entirely. Twenty percent switched to a competitor they believed was more transparent.The math here isn't complicated. A brand with a million customers could have lost 240,000 purchases in six months. Not because they did anything illegal. Because they weren't transparent about how AI was touching the customer experience.
The Adoption-Trust Gap
A June 2026 survey from Makeable found that 85% of managers use AI to understand their customers. But only 18% trust AI-generated insights more than direct customer research. Nearly one in five said their company had implemented an AI-generated recommendation that ended up negatively affecting the business.
This is the adoption-trust gap in action. Teams are deploying AI tools faster than they're developing the internal criteria to evaluate what those tools actually produce. Speed is winning. Validation is losing. And somewhere in between, bad recommendations are slipping through and becoming real decisions with real consequences.
IAB's 2026 research found that 83% of ad executives now say their company has deployed AI in the creative process, up from 60% in 2024. But a notable perception gap exists between what advertisers think consumers feel about AI-generated ads and what consumers actually feel, especially among Gen Z. Advertisers are bullish. Consumers are skeptical. And that disconnect is a trust liability hiding in plain sight.
What Actually Builds Trust
So what do we do with all this? The temptation is to either pump the brakes on AI entirely (not realistic) or to pretend the trust problem will solve itself as consumers "get used to it" (not happening).
The answer, I think, lies in something Glenn Sanford wrote about in a recent piece on scaling trust: "The ability to guide without micromanaging, to trust without abdicating, to know when to nudge and when to let it run, that's the same skill whether you're leading people or working with AI."
Trust isn't a feature you can bolt on. It's built through repeated delivery on promises, through accountability when things go wrong, through the kind of human judgment that no algorithm can replicate. AI handles the speed. Humans supply the discernment.
PwC's analysis of the personalization gap makes this point clearly: high-impact personalization requires coordinated maturity across technology, data, operating model, and measurement. Companies have more tools than ever, yet many can't produce meaningful outcomes because they've treated personalization as a tech problem when it's actually an orchestration problem.
The Real Metric
Here's my challenge to every marketing leader reading this: stop measuring your AI strategy by how much marketing it produces. Start measuring it by whether it makes the next customer interaction easier to earn.
That's a fundamentally different question. It forces you to think about the downstream consequences of every AI-generated touchpoint. It makes you ask: are we building trust, or are we just scaling promises we can't keep?
The 52% of consumers who said they'd pay more for a brand that's transparent about how it uses AI with their data? That's not a cost center. That's a revenue line. The companies that figure this out first won't just survive the trust gap. They'll turn it into a competitive advantage.
Marketing has always been about making promises. The best marketers have always known that the real work is keeping them. AI just made that truth impossible to ignore.
