Can you put your trust in Erica, Sandi, or Amy to take more responsibility for your finances without providing you with incorrect information or transferring money to the wrong place?
That’s the question posed by the Consumer Financial Protection Bureau in a study released Tuesday, in which the bureau expresses a number of worries about banks’ increasing use of chatbots to handle regular customer support inquiries.
The CFPB is concerned that banks or loan servicing companies may reduce their reliance on human customer service representatives and shift an increasing number of mundane jobs to AI. Furthermore, the agency claims that badly constructed chatbots may violate federal regulations governing how debts are collected or personal information is used.
“If firms poorly deploy these services, there’s a lot of risk for widespread customer harm,” said Rohit Chopra, director of the Consumer Financial Protection Bureau, in an interview with The Associated Press.
Banks have been managing an increasing number of customer service queries with chatbots for some years, generally with female-sounding names like Sandi for Santander, Amy for HSBC, or Eno for Capital One.
Under the Erica brand, Bank of America operates the largest and most successful financial chatbot. Erica, the final five letters in America, now handles hundreds of millions of inquiries from Bank of America clients each year.
Initially, bank chatbots were used to answer simple questions, such as altering a customer’s address or phone number or notifying a consumer where the nearest branch is or what the routing number on their account could be. However, as banks have invested millions in these services, chatbots have become increasingly smart, capable of understanding complete sentences and even assisting customers in moving money or paying bills.
According to the CFPB, almost four out of every ten Americans interacted with a bank chatbot last year, a proportion that is expected to rise.
Banks are preparing to launch increasingly powerful AI-like services. According to reports, JPMorgan Chase is working on plans to employ ChatGPT and artificial intelligence to help customers choose appropriate investments. Erica can be used by Bank of America bankers to create customer profiles and potentially recommend goods to those consumers.
While these services may save customers minutes or hours of waiting on hold to speak with a human to ask a routine question, the agency is concerned that these chatbots will be able to handle the nuances and complexities of consumer protection laws without providing customers with inaccurate information.
“There are federal laws that give customers rights when it comes to debts or disputing a transaction, and if customer service agents are being substituted for chatbots, that can pose problems if the software gives them inaccurate information,” Chopra explained.
The agency also discovered that senior clients or those whose first language is not English may become stuck in a customer support “loop” and be unable to reach a human agent.
“We found complaints of customers getting stuck in circular arguments with a robot or receiving incorrect information,” he said.