Quick Take
"We are repeatedly told that artificial intelligence will provide faster, more personalized and more efficient service, Stanford scholar Robert M. Kaplan writes after a series of chatbot misadventures. "Yet consumer frustration with AI interfaces has become commonplace, and many problems remain unresolved. "
Have something to say? Lookout welcomes letters to the editor, within our policies, from readers. Guidelines here.
Artificial intelligence is supposed to make life easier. This week, it did the opposite.
Over just a few days, I spent hours trapped in automated customer-service systems that could not solve simple problems and seemed determined to prevent me from speaking with anyone who could.
It began with a notice from a company contracted to pay me a small consulting fee for advising a public university. To process the payment, I needed to submit a W-9 form. I uploaded a completed form, but it was rejected because it had been signed the previous year.
I downloaded a new form, signed and dated it, and uploaded it again. The system acknowledged receipt and assured me that someone would get back to me shortly.
No one did.
Over the next two weeks, I resubmitted the form repeatedly. Each time, I received the same automated response: rejected. No explanation. No indication that a human being had reviewed the document.
Repeated requests to speak with a person went nowhere. The website offered no telephone number, and every inquiry generated another automated rejection. I spent hours corresponding with AI when a human being could likely have solved the problem in minutes.
The next problem started with an email that appeared to be from my bank, warning that my account might have been compromised. The message contained several hallmarks of a phishing scam, including an invitation to “click here” and enter account information.
Recognizing that clicking was probably a bad idea, I called the bank. After a lengthy hold, I was told to forward the suspicious email to the fraud department. I complied. All I wanted was reassurance that my account had not been hacked.
Soon afterward, I received an automated reply informing me that if I had clicked the fraudulent link, I should call the number on the back of my credit card. Since I had not clicked the link, the message concluded: “You will not receive any other response besides this automated email.”
My interpretation was simpler: “You’re on your own, buddy.”
The next frustrating encounter came when my wife and I purchased a washer and dryer for a rental property we own during a Costco sale. A helpful employee assisted us in placing the order. By the time we got home, however, the order had been canceled. No explanation. No instructions. No way to appeal.
The only option was to interact with a chatbot. After nearly an hour, we finally reached a human representative. The problem turned out to be absurdly simple: My Costco account listed my first name as Robert, while the order had been placed under Bob. That minor discrepancy was enough to trigger an automatic cancellation.
The next morning, we reordered the appliances only to see the order canceled again. This time, the delivery address for the rental unit did not match our home address. Over the next few days, the order was canceled several more times for similarly trivial reasons. Eventually, with the assistance of a human representative, we successfully placed the order. But by then the sale had ended, increasing the price by nearly $700.
My frustrations are hardly unique. Surveys consistently show that consumers prefer access to real people when problems arise. A 2024 study found that 94% of baby boomers highly valued companies that offer live phone interactions. Another survey found that 64% of consumers would rather companies not use AI for customer interactions. Another reported that 82% would prefer dealing with a human being, while only 18% were excited about using AI.
Yet businesses continue to rush toward automated customer-service systems because the technology reduces labor costs even when it leaves customers less satisfied.
We are repeatedly told that artificial intelligence will provide faster, more personalized and more efficient service. Yet consumer frustration with AI interfaces has become commonplace, and many problems remain unresolved. The issue is not simply that companies have inserted another barrier between customers and the help they need. Increasingly, they are shifting work from employees to consumers.
Customers must navigate automated menus, repeat information, troubleshoot problems and search for answers on their own. Tasks once performed by customer-service representatives are now delegated to the people seeking help.
To be fair, AI can perform some tasks remarkably well. It can summarize documents, draft correspondence, translate languages and answer routine questions. These capabilities are impressive and often genuinely useful. It is easy to understand why businesses are embracing AI: The technology can improve efficiency while reducing labor costs.
The problem is that customer service often seems to be an afterthought. In too many cases, AI is deployed less to improve service than to create a barrier between customers and the people who could actually help them.
The result is a system that saves businesses money while shifting costs onto customers. We spend more time navigating automated menus, repeating information and searching for a path to an actual human being. Delays increase, frustration grows and confidence erodes. In effect, businesses are imposing a hidden tax on their customers — paid not in dollars, but in time and effort.
Perhaps future AI systems will combine efficiency with genuine responsiveness. I hope they do. Until then, the most impressive customer-service innovation might be the oldest one: a company that answers its phone.
Robert M. Kaplan is a senior scholar at the Stanford University School of Medicine. He previously served as a professor and senior administrator at UCLA and UC San Diego and as a senior executive scientist at the National Institutes of Health.
