42% of consumers say they would pay extra just to talk to a human being.

Just for the privilege of speaking with someone who can listen, empathize, and use judgment. When “talk to a person” becomes a premium feature — like first-class legroom or a private banking advisor — we’ve crossed a line.

With my experience in business transformation space, I will be the first to tell you: I am not anti-AI. I have led initiatives that automated complex processes and improved cost efficiency while improving customer experience. But there is a difference between using technology to make experiences better and using it to make experiences cheaper. Right now, too many organizations are doing the latter and calling it innovation.

We are heading toward a two-tier customer experience economy. And if we are not deliberate about how we design it, we will end up in a world where empathy, attention, and human judgment are rationed by ability to pay.

The Bifurcation Is Already Here

Look at the data and the pattern is unmistakable. A 2025 SurveyMonkey study found that 79% of Americans strongly prefer interacting with a human over AI for customer service. Another report found that 50% of consumers would cancel a service entirely if it became solely AI-driven. And a Twilio study revealed that 83% of business leaders believe AI can replace human agents — while only 8% of consumers agree.

That’s not a gap. That’s a chasm.

On one side, executives see efficiency gains, deflection rates, and cost reduction. On the other, customers see the slow disappearance of the people who actually understood their problems. And the companies that are paying attention? They are turning human access into a product tier.

Airlines pioneered this years ago. First class was never just about the seat — it was about the dedicated gate agent, the concierge who remembers your name, who solves your problem without making you navigate a phone tree. In 2026, Southwest Airlines — the carrier built on egalitarian simplicity — is seriously considering adding first class for the first time in its history. Its CEO framed it as a response to industry-wide “premiumization.”

Banking is even more explicit. At one of the major banks, you get a digital robo-advisor at $0 invested but one can get a human advisor when they cross $250,000 of balance. At another bank, their premium tier members get a dedicated wealth management team, lounge access, and cultural event passes — but only if they maintain $200,000 in combined balances. The message is consistent across industries: human attention is expensive, so we will reserve it for customers who can afford it.

The Efficiency Trap

Here is the uncomfortable truth: the metrics most organizations use to evaluate their CX investments are actively misleading them.

A 2026 research found that 88% of customers said AI resolved their issue — but only 22% said the experience made them prefer the company. That’s a 66-point gap between “resolution” and “loyal.” If you’re a CX leader reporting deflection rates and cost-per-ticket to your board, everything looks green. But underneath those dashboards, customer relationships are hollowing out.

Gartner recently challenged one of the foundational assumptions of the entire AI-in-CX movement: that AI is cheaper than humans. Their prediction? By 2030, the cost per resolution for generative AI will exceed $3 — more than many offshore agents. Factor in data center costs, AI vendor price increases as the subsidy era ends, and the complexity of edge cases, and the economics of “replace humans with bots” start to look far less compelling.

This is what I call the Efficiency Trap. Companies race to Stage 2 of what I have mapped as the Service Evolution Model — they move from viewing service as a Cost Center (Stage 1) to treating it as an Efficiency Engine (Stage 2), where the primary goal is reducing cost per ticket. And that is where they stall.

The companies that break through to Stage 3 — the Relationship Hub — understand that service isn’t overhead to minimize. It’s where loyalty is earned, where churn is prevented, and where lifetime value is built. They use AI to augment human capability, not replace it. And the very best — Stage 4, Trust Architecture — they make human access the product itself.

The Equity Question Nobody’s Asking

Here’s where this conversation gets uncomfortable, and where I think the CX and Service Design profession needs to do some real soul-searching.

We are making a design choice when we design 2-tier service architecture, one that is premium human support for high-value customers and another for everyone else that is bot driven. We are saying that empathy, patience, and human judgment have a price point. That the customer with $200,000 deserves to be heard by a person, and the customer with $2,000 gets a chatbot.

I come from architecture and industrial design before I came to CX. In those fields, there is a concept of ‘universal design’ — the idea that the best-designed environments work for everyone, not just the privileged few. A well-designed building doesn’t put the ramp in the back. A well-designed product doesn’t hide its best features behind a paywall.

Yet that is exactly what is happening in customer experience. We are putting the ramp in the back.

This matters especially in industries where customers don’t have a choice — healthcare, utilities, financial services, telecom. When your internet goes out during a work-from-home day, or your insurance claim is denied, or your elderly parent can’t navigate a chatbot flow, the ability to reach a human isn’t a luxury. It’s a necessity.

Building the Bridge: From Efficiency Engine to Trust Architecture

I am not arguing against AI in customer experience. I am arguing against thoughtless AI in customer experience. The question isn’t “should we automate?” It is “what are we automating for?”

Three principles for leaders who want to get this right:

Design for the floor, not the ceiling. Instead of asking “what is the minimum viable experience for our lowest-tier customer?”, ask “what’s the baseline level (table stakes) of human access that every customer deserves?” Then use AI to make that baseline affordable rather than eliminating it.

Measure what matters. Deflection rates and cost-per-ticket are operational metrics, not experience metrics. Track relationship indicators: repeat purchases after a service interaction, NPS shift post-resolution, customer effort and emotional satisfaction. If your AI is resolving tickets but customers feel nothing, your metrics are lying to you.

Make the handoff seamless, not invisible. The best implementations I have seen don’t try to hide the AI. They use it transparently — “I am an AI assistant, and I can help with X. For anything more complex, I will connect you with a person.” Customers don’t resent AI. They resent being trapped in it the endless loop that these AI bots create.

The Choice Ahead

A Medallia executive recently coined the term “AI tax” — the idea that as more experiences shift to AI, customers may be less forgiving of AI errors than human ones, creating a hidden cost in brand equity and loyalty that doesn’t show up on any dashboard.

CX leaders are making Technology, staffing and service design architectural decisions right now that will define the customer experience for the next decade. The question isn’t whether AI will transform service. It will. The question is whether we will design that transformation with the same care for the human being on the other end that we bring to the algorithm.

It is a bet that the companies that win the next decade won’t be the ones that most aggressively automated their customers away from people. They will be the ones that figured out how to make human connection scalable, accessible, and present at every tier.