From The Economist's 1843 Magazine, February 2:
AI is poised to eliminate customer-service agents. We’ll miss them when they’re gone
I was only trying to order my contact lenses. A mundane task, undertaken nose-in-phone while walking somewhere, probably late, likely crossing a road, which is my preferred, death-adjacent method for all personal administration. Why look at buildings and trees, at passing humans and oncoming vehicles, when you could be getting stuff done on your phone?
Vision Direct, one of Britain’s most popular online contact-lens stores, promised I could order my lenses in a seamless and rapid process that would involve three clicks and, in its ideal form, zero interaction with another human. Which was precisely what I wanted. I interact plenty. I do not need to interact when I’m ordering contact lenses.
And yet, with a certain inevitability after such a promise, the payment didn’t go through. I clicked again. Nope. Then I went back to the beginning of the much-vaunted three-click process. Nope. Three clicks now felt more like a taunt. I’d done at least eight. After my fourth failed attempt, by which point I figured I’d paid quadruple the appropriate amount for my contact lenses and yet still wasn’t going to receive them, I felt my chest tighten. Oh please, God, no. I was going to have to call.
I never want to call. Who does? People like my mother, maybe. She thinks nothing of getting straight on the phone to a company, as if there might be a dedicated person sitting right at the source of the gas or the broadband or the mobile signal, able to resolve the problem instantly. She is stunned and infuriated when this doesn’t happen, and will pass entire mornings on the phone, waiting, explaining, being transferred, explaining again, gradually discovering new dimensions to her rage.
But most of us know the truth. Call the customer-service number and you enter the underworld. Finding the number is hard enough. Dig it out of the dankest corners of a website – because God knows they don’t want you to call either – and then you’re on hold, in the queue, listening to the music, waiting, hearing the music loop back to the beginning, waiting a bit more. By the time the agent finally gets on the phone, your irritation can’t help but spill over a little, so you apologise even though it’s not your fault, just as it’s not theirs, but then whose fault is it? And why do you never get to talk to whomever is actually to blame?
In the best-case scenario, your problem is resolved and you will probably have been a little curt with someone who doesn’t deserve it. The worst-case scenario – interminable waiting and lack of resolution – is a day-ruiner. Recently, a friend told me she’d actually cried on the phone to her broadband company. The agent tried to make her feel better. “It’s only broadband!” But it’s not only broadband, is it? It’s the distinct sensation of life slipping between your fingers, the feeling that you are lost in a dehumanised process enacted by a giant machine which doesn’t care about either its customers or its employees. It is automated hell. Still, the customer-service industry now believes it has hit on a technological solution that will lead us serenely out of the labyrinth. Or so they say.
For centuries, all transactions were face-to-face: at the market, over the counter, in the shop. If there was a problem, you walked in and had a conversation. In the late 19th century telephone switchboards opened a new avenue of communication. As more households had their own phones, the telephone call became the main conduit between company and customer. By the 1960s the volume of calls required their own infrastructure: the call centre.
In the decades since, the numbers of people working in call centres have steadily grown. Over 800,000 in Britain, nearly 3m in the United States, approximately 17m globally. Max Ball, an analyst at Forrester, a consultancy, told me that the call centre is usually perceived within a company as a “cost centre” – a part of the business that eats up resources without adding to the bottom line. According to a recent report by the Institute of Customer Service (ICS), customer service was costing British businesses £7.1bn a month as their employees spent, on average, over half a day a week dealing with problems.
Turnover in the customer-service sector is fast. One manager said if they kept someone for two years, they were lucky. Agents get tired of the abuse and blame from customers for things that were not their fault. A recent survey of customer-service employees by the ICS revealed that over a third had experienced hostility in the last six months, and a third of those said this happened so frequently that there was no point in reporting it to their managers.
Call centres require a large workforce, intensive training and are plagued by inefficiencies, mostly measured in time: how long a customer has to wait on hold, how long a call runs until the issue is resolved. As such, the challenge of any company with a call centre is to reduce that time. “If I shave a minute off every conversation,” said Ball, “that’s hundreds of millions of dollars every year that I’m saving.”
Attempts to time-shave and cost-save, both infrastructural and technological, have been numerous. There was Interactive Voice Response, a system that asks callers to choose between options on their keypad. In the 1990s speech recognition, whereby a computer program responds to a customer’s request, was supposed to reduce the need for agents. But a human was still required to mop up the mess after a customer had spent five minutes shouting into their phone at an uncomprehending machine. Outsourcing – largely to India and the Philippines – was another cost-reducer. The Philippines is now the call-centre capital of the world, with over 1.5m people employed in the industry. Since the pandemic, like call-centre employees everywhere, they mostly work from home.
Now the sector faces its next fundamental shift, one which looms over every industry – the advent of generative AI, particularly chatbots built on large-language models which can communicate almost with the fluency of humans. Of the various views, which run from the utopian to the panicked, about the ways in which generative AI will transform the world and work, there is broad consensus that customer service is among the first in line for an overhaul. Gartner, a consultancy, recently predicted that, by 2026, investment in generative AI will have reduced the number of customer-service agents by 20-30%....
....MUCH MORE
Earlier today: