Well, it happened. AI fooled the AI analyst during a recent customer interaction. Not intentionally, and not in a bad way. But it clearly underscores some of the shifts Metrigy is seeing in our latest research, as well.
My story involves my months-long search for a Mother-of-the-Bride dress. The variety of options online are much better than going into stores, so like many of my friends in this era, I ordered (too many) dresses, tried them on at home, and returned those I didn’t want. After about 24 dresses, I found one I liked but noticed it was now on sale.
I could have returned it with the others I had bought, and just repurchased the same one sale, but I figured I would email and ask about the situation. My email identified the dress, my order number, and the fact that it was now $100 cheaper. I asked if they would credit the difference or if I should return it and order a new one.
Just minutes later, I received an email response that they had issued the credit. No arguments. No problems. Just a pleasant response and my issue was resolved—so much that I even replied thanking them for the fast response and good service and then filled out the satisfaction survey with a 5-star score for Alicia.
A week later, as is my prerogative, I changed my mind and wanted to keep a different dress in this order and return the one for which they issued a credit. So, I sent another email explaining my change and asked how to handle the return that I had already submitted. My response was similar in tone to the last one, but it said she needed to turn it over to a human agent.
Signed… Alicia, the store’s virtual assistant.
BAM! I was fooled. And it brought my mind right to a chart I had just created for Metrigy’s CX Optimization 2025-26 global research study of 656 companies about the importance of people vs. technology.
In response to the question: “Though both are important, which is most important to delivering successful customer interactions: people or technology,” 72% of companies said it was people in 2024. Within one year, that number has dropped to 59%. True, the majority still think people are more important, but the margin dropped considerably in one year.
What is even more alarming is what the Metrigy Research Success Group said. (The Success Group has at- or above-average business metric improvements in multiple areas using AI in CX.) Among this group, a whopping 58% said technology is more important than people. One year ago, only 45% of the Success Group selected technology over people.
AI is getting good— really good —in many customer service situations. The fact that technology is more important than people among the Success Group does not suggest technology is a replacement for people in all cases. In fact, technology is what’s making people as good as they are in live interactions as the consensus that emerged during interviews with research participants illustrates:
Technology and people are both important.
In some cases, technology can address a customer service issue without real-time human involvement.
Humans rarely address customer service issues without technology involvement.
Humans are valuable, but AI and other CX technologies make them better, smarter, faster, and more successful.
Therefore, the value technology brings to humans makes it more important overall.
To put this shift in perspective, Metrigy estimates that currently 31% of all customer interactions are resolved without human involvement, and we expect that to increase to 45% by 2027 and 60% by 2029.
In my experience, the AI agent was surprisingly good and fast. I was not expecting a response that quickly, nor was I expecting resolution without some argument. Once a human did get involved, the response took a full day for resolution. So, the technology-only resolution was faster but just as polite as the human resolution.
The speed of response is the top reason consumers opt to use AI agents when interacting with businesses: 57% of consumers say they will use an AI agent when it saves them time.
Generative AI’s increasingly proficient ability to support conversational interactions, both over voice and text, is driving more consumers to select AI agents. But in some cases, like my own described here, AI agents are jumping in without the customer selecting to use them.
If I’d had the opportunity to select an AI agent or human agent, I would have selected a human agent. After this experience, though, I’ll give AI agents a chance. Why? It was fast, flawless, and caused no friction whatsoever.
For the foreseeable future, AI and humans will work together. Some issues will be best served by AI agents and others by human agents. But one thing that is clear: 86% of companies say AI and human agents are working together to serve customers.
The top three ways AI is helping human agents include:
Automating tasks for follow-up (52%),
Gathering information for humans (50%), and
Suggesting responses for humans (47%).
Nearly 60% of companies measure the difference between AI and human agents, as well. For now, 58% say human agents are better at addressing customer issues than AI agents, which aligns well with the aforementioned “people vs. technology” question. And not surprisingly, 63% of Metrigy’s Research Success Group already says AI agents are better at addressing customer issues.
As AI continues to improve, others will follow the Success Group sentiment, and humans’ roles will shift to circumstances they’re much better qualified to handle. The top issue is sales and upsell. Perspectives also are driven by the ages of customers, the type of company, and the type of interaction. But the bar continues to move as to what AI can and cannot handle.