AI adoption backfires as frustrated customers turn away from brands

The shoes appeared sharp and detailed, while the model's movements looked natural. But on closer inspection, Ha noticed that a pattern on the right shoe suddenly appeared on the left as the model walked, exposing the footage as artificial.

"I watch these videos to see the true details and how the shoes actually look in real life so I know if they suit me," she says. "AI videos are always flawless, making them impossible to trust."

Instead of buying from the site, she then chose retailers that feature real people trying on footwear, determined to get an authentic feel for what she was purchasing.

Nguyen Hoang Nam, 35, of Ho Chi Minh City, has similarly lost his temper after dealing with automated systems.

While driving one afternoon, Nam received a phone call from an electronics superstore where he had just shopped. Assuming it was an urgent matter, he pulled over to take the call, only to hear a robotic voice asking him to complete a customer satisfaction survey.

Annoyed by the intrusive call, he dialed the store's customer hotline to complain, only to be bounced through several automated menus before reaching a real person.

"It is the same everywhere," he laments. "Companies keep telling you to message an AI chatbot or dial an automated helpline for faster assistance, but nothing ever gets solved."

For him, the issue is not that businesses use modern technology, but that automation is turning straightforward interactions into convoluted, time-draining ordeals.

An AI-generated product video on an e-commerce platform in October 2026. Photo by Luu Quy

Setting boundaries for AI

The frustrations experienced by Ha and Nam highlight the risks of deploying AI across customer-facing touchpoints.

While companies adopt automation to slash overheads, increase efficiency, and upgrade service, the strategy backfires when it prevents buyers from verifying information or adds friction to resolving everyday issues.

A study released by market research firm Gartner in March revealed that 50% of consumers prefer brands that avoid generative AI in customer-facing materials like marketing campaigns, advertisements, and public messaging.

Furthermore, 68% of respondents said they regularly question whether the content and information they encounter is genuine.

Another Gartner survey found that while half of participants felt interacting with companies had become easier with generative tools, 87% said businesses must provide an option to speak with a human. Only 27% expressed willingness to give a chatbot another chance after once having a negative experience.

Similar skepticism has surfaced in customer support. A customer experience survey of adults in the United States and Canada done in February by research advisory firm Metrigy found that 84.9% favored interacting with human staff over AI agents.

Even when assured that both a human and a bot could resolve their problem equally well, 80.1% still opted for a human.

After herself struggling with AI chatbots, Minh Anh, who runs an online boutique in the northern Phu Tho Province, decided against turning customer communication over to automation entirely.

Testing an automated solution earlier this year, she found the bot swift and knowledgeable about standard operations.

"Most shoppers reaching out to our store have unique requests rather than generic questions," she explains. "The bot ended up creating more problems because it misread their intent, gave bad advice, and left customers frustrated."

According to Vo Thi Diem Trang, who teaches professional communication at RMIT University Vietnam, users walk away from AI-powered services not simply because the software lacks intelligence, but because it feels artificial.

In a study published in October, she highlighted several pain points that drive consumers away from AI applications. Chief among them is the impression that service providers are prioritizing company profits over customer convenience.

Confidence in AI tools also plummets when systems lag, freeze, or suffer technical glitches. Contradictory answers, poorly explained recommendations, and guidance out of touch with local realities has the same effect.

Data privacy poses another major hurdle. Trang says that when users doubt the security of their personal information, particularly regarding sensitive matters such as health or physical appearance, they are quick to abandon the service.

Based on these findings, Trang suggests that businesses should rethink how they define a successful AI deployment.

Instead of merely chasing higher processing power or algorithmic cleverness, companies must ensure the technology delivers real value to users, she says. "Businesses should focus less on making AI smarter and more on making it authentic. Having a responsive team of real people available remains the ultimate hallmark of authenticity."

An attendee interacts with a virtual assistant during a technology exhibition in Hanoi in September 2026. Photo by Trong Dat

This mirrors guidance from Gartner. Eric Keller, a senior research director at the company, says enterprises should not force customers through AI funnels in every scenario.

Automation should be reserved for touchpoints where it reliably solves problems, backed by a seamless escalation path to human staff whenever the system encounters ambiguity, he says.

The ultimate goal is not trapping every inquiry inside a chatbot, but guiding customers along the quickest, most effective route to a solution, whether through automation, a live agent, or a combination of both, he says.

Companies should therefore stop measuring AI success strictly by call deflection numbers or automated ticket volumes, and instead focus on whether customers truly get their issues resolved and feel inclined to return, he adds.