Nearly ten years ago, Armen Kirakosian began working as a calls center agent. He recalls the hassles of his first job, including the irate clients, the need to manually record notes for every call he handled, and the continual menu scanning for information.
The Athens, Greece, 29-year-old is no longer taking notes or selecting from a myriad of choices because of artificial intelligence. He frequently has whole client profiles in front of him when someone calls, and he may be able to identify the issue before the consumer even says “hello.” He has more time to devote genuinely assisting the client.
According to Kirakosian, “A.I. has taken (the) robot out of us.”
About 3 million Americans work in call centers, and millions more do the same in call centers across the globe, where they handle billions of questions annually about anything from malfunctioning iPhones to shoe purchases. Kirakosian is employed by TTEC, a company that offers third-party customer service lines in 22 nations to companies in sectors like banking and automobiles who want additional capacity or have outsourced the call center operations.
These calls might be unappreciated labor. McKinsey estimates that within a year, around half of all customer service representatives resign, with many citing boredom and stress as causes.
The industry term for a lot of what these agents handle is “break/fix,” which indicates that something is incorrect, confused, or broken, and the client wants the person on the phone to solve it. Who will be responsible for fixing it—a human, a computer, or a human augmented by a computer?
Already, AI bots are taking over increasingly ordinary call center work. Some employment have been lost, and the future job market for these people is expected to be bleak, with losses ranging from modest single-digit percentage points to up to half of all call center positions disappearing over the next decade. The decline is unlikely to equal the more terrible estimates, however, because it has become clear that the sector will continue to require humans, possibly with even greater levels of learning and training, as some customer service issues grow increasingly difficult to resolve.
Some financial institutions have previously experimented substantially using AI to address customer service challenges.
By 2024, the Swedish buy-now-pay-later startup Klarna had replaced 700 of its about 3,000 customer care representatives with chatbots and artificial intelligence. The outcomes were uneven. Klarna discovered that even while the firm saved money, in some situations—like complex identity theft issues—higher experienced human agents were still required. Seven internal freelancers were employed by Klarna earlier this year to deal with these problems.
Klarna rehired a small number of customer support representatives earlier this year, conceding that certain concerns, like as identity theft, could not be handled as successfully by AI.
According to Gadi Shamia of Replicant, an AI software company that makes chatbots sound more human, their vision of an AI-first contact center—where AI agents handle the majority of conversations and fewer, better-trained, and better-paid human agents support only the most complex tasks—is rapidly coming around, according to consultants at McKinsey.
Although it has improved, the call center client experience is still far from ideal.
Interactive voice response systems, or IVRs as they are called in the industry, have long been used to handle the initial customer support contact. When instructed to “press one for sales, press two for support, and press five for billing,” customers engage with the IVR. The 2010s saw an upgrade to these antiquated methods, allowing users to trigger the system with words like “sales,” “support,” or straightforward expressions like “I’d like to pay a bill” rather than wading through a confusing array of menu alternatives.
Customers click the zero button on their keypad in the hopes of speaking with a human, a process known as “zero out,” since they are impatient with these menus. After a consumer “zeros out,” it’s also not unusual for them to be placed on hold and transferred since their request was not fulfilled.
Democratic Senator Ruben Gallego of Arizona and Republican Senator Jim Justice of West Virginia have introduced the “Keep Call Centers in America Act,” which would mandate clear methods of contacting a human agent and offer incentives to businesses that maintain call center employment in the United States. They are aware of the general dissatisfaction that Americans have with IVR.
Businesses are attempting to implement phone systems that can forecast where to send a client without requiring them to navigate a menu and have a general understanding of customer care requirements. The “ChatGPT Agent” service, developed by OpenAI, the company that created ChatGPT, will be available to customers who can comprehend sentences like “I need to find a hotel for a wedding next year, please give me options for clothing and gifts.”
Bank of America claims that after introducing “Erica,” their chatbot in 2018, it has been more successful in incorporating these characteristics. The agent directs the client to the appropriate department when Erica is unable to address their request. In addition to being analytical and predictive, Erica is now aware that a client may have many subscriptions to the same service or frequently have a low balance, in which case they may want greater budgeting assistance.
According to a statement released by Bank of America this month, Erica has been utilized 3 billion times since its inception and is handling an increasing number of customer support queries. This chatbot’s name is derived from the final five letters of the company’s name.
As vice president of product and innovation at TTEC, James Bednar has devoted a significant portion of his professional life to reducing the unpleasantness of customer support calls for both the organization and the caller. According to him, these technologies may someday completely eradicate IVR, negating the need for anybody to “zero out.”
According to Bednar, we’re approaching the point when AI will direct you to the appropriate individual for your issue without requiring you to navigate through those options.
