Why it's not going to send your customers running for the hills.
6 minute read
There's a persistent narrative in our industry (and in the general discourse) that people don't want to talk to AI, especially about AI and voice. But the data tells a very different story. Customers are not anti-AI as much as they are against turning one cause of frustration into multiple ones. People care way less about who solves their problem; they care way more about whether the problem gets solved. Period.
As long as voice AI behaves the way a real conversation should, people are more open to it than our common discourse would have us assume.
According to the Metrigy Research 2025 consumer survey, the vast majority of customers are comfortable with voice AI agents when two basic expectations are met: either their issue is resolved, or they can seamlessly escalate to a human when it isn't.
Here are the numbers:
More than 80% of customers are open to voice AI, provided it delivers results rather than dead ends. That runs completely counter to the prevailing narrative that consumers are inherently resistant to AI and automation.
This openness to voice AI shouldn't be surprising when you look at how customers actually engage with contact centers. In the same Metrigy survey, 82% of all contact center interactions involve voice conversations, either as the initial channel or as an escalation point. That's up five points from 2024.
Even when customers start somewhere else, such as chat, messaging, or self-service, they still end up on a call when things matter. Voice is a critical safety net for escalation (and resolution). It's where nuance gets its due, and where trust is built.
More specifically, voice AI interactions have grown 7x year over year, now representing over 75% of all AI-led interactions. This validates that voice AI isn't a niche experiment. It sits directly in the critical path of how we can improve customer experience.
When they need help, customers don't dwell too much on the philosophy of where, how, and why you're using AI. Their primary motivation is practicality. They're saying, “Help me solve the problem, and don't trap me in a Catch-22-style loop if it gets complicated.”
Voice can convey context and emotion in a way other channels simply can't. It's faster than typing, easier than navigating menus, and more natural for explaining complex issues. That's why customers are willing to use voice AI earlier than many expected. Especially if they can articulate those issues in their native language.
That's a frustration that has nothing to do with AI and everything to do with being understood. A contact center that can speak 100 languages (and get the accents and inflections right) removes that barrier before the conversation even starts. Talk to it in Spanish, French, Tagalog, or whatever language they're most comfortable in. And it meets them there instantly, no transfer to a language line, no wait for the one bilingual agent on shift. For a huge share of customers, that's the difference between feeling like an afterthought and feeling understood from the first word.
But that same bandwidth raises expectations. When voice AI fails, it fails rather embarrassingly. Poor transcription, weak intent recognition, or broken handoffs are that much more noticeable when they're happening live. Complete model crashouts and AI-generated nonsense are not unheard of either, and they're great at demolishing trust.
Finally, emotional care shouldn't be discounted. A recent report found that over two-thirds of people are likely to trust AI agents that behave like human agents and demonstrate empathy and a willingness to resolve issues. But many AI agents are built with goals that overpower the emotional ones. They're primed to prioritize tickets closed, conversations per hour, or issues marked “resolved” without involving a human agent.
The person on the other end can sense when they're being rushed for the sake of the bottom line. Just like you can sense that a server at a busy restaurant is trying to rush you off a table so they can seat the next diners and increase turnover for the evening. It puts a real damper on the entire experience.
One of the most important signals in the Metrigy data is how explicitly customers value escalation. The 27.2% who demand a human option are not necessarily anti-AI. They're signaling that they want resolution (and fast). They expect the system they're dealing with, be it human-led or AI, to recognize complexity and respond appropriately.
A common issue is that AI systems are assigned the task of containment rather than escalation. Escalation is seen as a failure of the AI system to resolve the problem, and so instead the AI loops and loops over itself, trying to keep the customer in its purview. The customer, as a result, is trapped in an AI limbo that promises no escape.
When these systems are designed, escalation must be considered as a possibility from the start:
This is where many voice AI deployments fall apart. The bot lives in one system, and the agent lives in another. When those systems don't connect, all the context behind escalation gets dropped, and customers pay the price.
I (and probably most readers) can speak from personal experience. I recently had to reauthenticate and repeat my problem numerous times when I was on the phone with a major US financial services company. It's a great example of how integration failures, not an AI itself, are the lead drivers of poor experiences.
Customers have already told us what they expect. They're open to voice AI. They just want it to respect their time and solve the problem.
The evidence is everywhere:
Today, voice is the infrastructure that powers experiences. Our CPaaS team's piece on voice technology is a great examination of how voice no longer lives in a silo. It's embedded inside the workflows people already use, working practical magic in the background:
In each case, voice is a crucial and often invisible ingredient. Like salt in a recipe, when it's used right, you don't notice, but everything tastes better.
CEO, 8x8
Samuel Wilson is Chief Executive Officer at 8x8, bringing 25+ years of experience across finance, investment management, and sales, plus deep company leadership spanning Chief Financial Officer, Chief Customer Officer, and go-to-market roles to drive disciplined growth and customer outcomes.
Stay current on what matters in CX and IT. Subscribe to our LinkedIn newsletter for regular analysis on the decisions shaping customer experience, technology, and AI. Clarity you can act on.
