Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
How many times do you call a customer service help desk before reaching breaking point? I came close when trying to arrange an internet connection for my new home earlier this year. Each time we booked an installation, we subsequently received an automated text message postponing our appointment. The process was repeated over weeks in an infuriating doom loop. “I don’t see this in our records,” one call centre adviser said over and over again in a scripted response. All I craved was a human reaction acknowledging my frustration. Instead, I was asked to fill out a satisfaction survey.
Customer service is often so awful that it’s hard not to think that artificial intelligence tools will make it better — from faster responses and tailored feedback to predictive analytics and smart routing. But I’m not sure how much of a panacea it will be. AI is taking hold in this space partly because so many of these processes have been dehumanised. It is patching over a problem technology itself helped create.
Belinda Parmar, chief executive of corporate advisory company The Empathy Business, says AI is helping to fix a “broken system” created by cost-cutting, automation and an obsession with efficiency. “Call centres were set up to manage processes, not people,” she says. Their language is entirely transactional — from abandon rates to average handling times — and designed “to dehumanise the operating environment”. An “agent” on their 60th call of the day, who is penalised for deviating from a script, is hardly in a position to exude warmth.
Some experts say that AI can indeed help on this issue and boost customer satisfaction and loyalty, including in interactions with chatbots or emails. AI can already be used to recognise emotional cues, analyse text or voice tone to sense anger or sadness. If it detects distress, AI programmes can flag the situation to a supervisor or alert an adviser before the situation escalates. Scripts can adapt in real time.
Technology companies argue AI tools, or agents, should handle tedious and repetitive tasks to free up human workers to do the more meaningful work. The idea is to enhance customer service efficiency while preserving the essential human element. Some companies are attempting to strike this balance. Octopus Energy in the UK, for example, employs generative AI to transcribe calls, summarise conversations and draft automatic email responses, leaving workers to focus on complex customer cases. Yum Brand’s Taco Bell has AI handling drive-through orders, so employees can focus on more nuanced interactions.
But last year, K Krithivasan, chief executive of India’s largest IT company Tata Consultancy Services, told the Financial Times that wider adoption of generative AI among large companies might make most call centres obsolete if businesses could predict calls before they happened and proactively resolved issues.
This approach would make many happy. Good customer service is increasingly perceived as frictionless. From entering a gym to getting on a bus — the tap of a smartphone is often all that is needed. Conversation is optional.
Yet even as human interaction is shrinking in some realms, it still matters and not just for those that can pay a premium for personal attention, such as in luxury retail.
Despite AI’s rapid improvements, new research from Harvard Business School shows we still value human empathy more. Nearly 3,500 participants were asked to share emotional experiences and were then shown identical empathetic responses, some labelled as human-written and others as AI-generated. Those who thought a human wrote the response felt more positive, supported and satisfied. Notably, 30-50 per cent were willing to wait hours — or even far longer — for a human response over an instant AI one.
Too often, so-called “corporate empathy” feels fake. The endless pursuit for the perfect response sacrifices authenticity — like the hyper media-trained chief executive whose words are polished to the point of lifelessness. While AI can personalise customer communication, it risks being intrusive rather than truly empathetic. Would the recent email sent to me from Netflix about the increase in subscription price have landed better if it acknowledged my bad mood that week based on my binge-watching habits? Possibly. Would I trust it more? Unlikely.
Workers in customer-facing roles need to be supported and rewarded for compassion and natural responses to real-life scenarios over efficiency. As Parmar points out, the goal of AI should be to “enhance human connection, not replace it”.
anjli.raval@ft.com