From hotels to retailers, the lion's share of businesses are deploying large language models, with top use cases across search and customer support. But these models aren’t perfect.
Consumers are frustrated with AI chatbots that can’t answer their inquiries, and patience is growing thin.
Some consumers will simply give up, according to a Parloa study released Monday. Three in five will repeat themselves once before abandoning automated support.
These points of friction can prove costly. Here are three numbers on the cost of AI failing customers:
50%
The amount of consumers who blame business leadership when its AI support fails to answer a customer’s query according to a Kim.cc survey of 1,000 U.S. consumers released last week.
When consumers are faced with an AI chatbot or tool that can’t solve their inquiry, consumers aren’t forgiving.
“The big takeaway for leaders is that customers do not see bad AI support as a tech issue. They see it as a company problem,” Sachin Jaiswal, co-founder and CEO of Kim.cc, said in an email to CX Dive.
“If a customer gets a wrong answer, gets stuck or feels like the company is hiding behind a bot to save money, the loss is much bigger than one bad ticket,” Jaiswal said. “It can lead to loss of trust, loss of brand reputation, and eventually, loss of that customer.”
The all-too-frequent friction caused by AI is likely why over one-third of respondents reported feeling immediately frustrated when customer support is powered by AI, according to the Kim.cc survey.
91%
The accuracy of Google’s AI model, according to an analysis by AI startup Oumi for the New York Times earlier this spring.
Google processes more than 5 trillion searches a year, meaning that with a 91% accuracy rate it provides billions of wrong answers a year. One of those wrong answers recently made headlines. Earlier this month, a German regional court issued a temporary injunction against Google over false information generated by its AI Overviews feature. Two companies had brought the case after certain searches led Google AI Overviews to falsely tie the companies to scams and shady business practices. The ruling said the false and defamatory statements generated by its AI Overviews feature is the company’s own content, and as such would be liable for it. While Google is fighting the ruling, the tech giant is currently set to cover 80% of the legal costs. If the ruling stands, it will open the door for more lawsuits and defamation cases against tech companies leaning on generative AI
$5,000
The amount it cost an autodealership whose rogue AI chatbot said a car was worth more than it was.
When a Toronto man decided he wanted to sell his car, he submitted an inquiry to the BMW dealership from which he bought it, CBC News reported last month. He received a text message from “Quinn” who made a buy-back offer of 27,162.79 Canadian dollars ($19,119). It turns out Quinn was an AI chatbot, and the offer was made in error. The car dealership walked back the offer, until CBC News contacted BMW Toronto for comment. The dealership reinstated the original offer, paying the man about CA$7,000 Canadian dollars (about $5,000) more than what the dealership actually thought the car was worth.
