46% of customers have lied to AI so they could talk to a human being
Shane Schick tells stories that help people innovate, and to…
Nearly half of U.S. consumers have either exaggerated details about an issue or provided inaccurate information to automated artificial intelligence (AI) tools so they could connect with a service rep, according to research published by Telus Digital.
The Vancouver-based technology services provider conducted an online poll of approximately 1,000 American shoppers to produce its findings, which were summarized in a press release.
Besides not telling AI service tools the truth, 89 per cent of those surveyed said they have used some kind of workaround when their service experience fell short.
Among the challenges consumers face, nearly a quarter cited explaining their issue in a way an AI tool would understand them. Repeating the same information was another common problem among 22 per cent.
The workarounds, meanwhile, included rewriting their question, cited by 47 per cent, and 28 per cent opted to simply start over with a new session. As for those who resorted to lying, 33 per cent said they have done so more than once.
“Keeping customers and relieving frustration calls for AI that understands their requests and resolves their issues, alongside human agents who are easy to reach and have real-time guidance and context when a customer needs a person,” Telus said in a quote accompanying its research findings.
360 Magazine Insight
Telus Digital was more charitable in describing what consumers are doing as “bending the truth,” but these exaggerations and inaccuracies were clearly acts of deliberate misinformation.
This speaks to the growing sense that AI-powered customer service may be more of a detriment to loyalty than a way to enhance it. For instance, 77 per cent of those surveyed said a guaranteed human support option would make them more likely to choose a brand for everyday purchases. Conversely, 30 per cent said they abandoned a purchase when they learned AI was the primary support mechanism.
Workarounds like lying should be particularly concerning to customer experience (CX) leaders making investments in AI because the whole premise of these tools is that they will continuously learn and improve. There has probably been an assumption that the data used to train AI data will based on truthful information. Over time, the impact of inaccurate or exaggerated statements could degrade the technology’s performance even more over time.
This trend towards lying was even higher (57 per cent) among the 18- to 29-year-old cohort, suggesting that AI is hampering a brand’s ability to build trust among the customers they should be trying to keep.
The most actionable takeaway is obvious: before deploying AI, CX teams need to test service workflows as though they were customers, using everyday language and scenarios that could trip up the tools. Spotting the points where AI-powered service breaks down may be the only way to stop the lies and start seeing real return-on-investment (ROI) in the technology.
Shane Schick tells stories that help people innovate, and to manage the change innovation brings. He is the former Editor-in-Chief of Marketing magazine and has also been Vice-President, Content & Community (Editor-in-Chief), at IT World Canada, a technology columnist with the Globe and Mail and Yahoo Canada and is the founding editor of ITBusiness.ca. Shane has been recognized for journalistic excellence by the Canadian Advanced Technology Alliance and the Canadian Online Publishing Awards.







