Last week, London caught me off guard in a way that perfectly captured the AI paradox in customer service.
I had just wrapped up hosting the 𝗖𝗦𝗦 𝗘𝘅𝗲𝗰𝘂𝘁𝗶𝘃𝗲 𝗙𝗼𝗿𝘂𝗺, an event steeped in discussions of AI’s expanding role in customer service and support, and was heading toward Covent Garden for dinner with peers when I realised I might need some cash.
My Apple Wallet card works seamlessly in the States, but the London ATM had other plans:
𝘗𝘩𝘺𝘴𝘪𝘤𝘢𝘭 𝘤𝘢𝘳𝘥 𝘳𝘦𝘲𝘶𝘪𝘳𝘦𝘥.
No card. No cash. Just my phone. I dialed the bank’s international support line.
The rep sounded flawless—perfect cadence, zero hesitation.
“𝘜𝘮… 𝘢𝘳𝘦 𝘺𝘰𝘶 𝘢 𝘳𝘦𝘢𝘭 𝘱𝘦𝘳𝘴𝘰𝘯?”
"𝘐 𝘢𝘴𝘴𝘶𝘳𝘦 𝘺𝘰𝘶 — 𝘐'𝘮 𝘳𝘦𝘢𝘭," she replied, then interrupted my follow-up question before I could finish it. That interruption convinced me she was human; AI still waits its turn in conversations.
𝗧𝗵𝗲 𝗧𝘂𝗿𝗶𝗻𝗴 𝗧𝗿𝗮𝗽
AI hasn't replaced agents, but it touches everything — call routing, noise filtering, accent smoothing, and more.
We've optimized voices to perfection, then doubt them for sounding perfect.
We train agents with machine-generated scripts, then question their authenticity when they sound... machine-optimized.
𝗧𝗵𝗲 𝗡𝗲𝘄 𝗛𝘂𝗺𝗮𝗻 𝗖𝗔𝗣𝗧𝗖𝗛𝗔
To convince skeptics, agents now cough, laugh, recite anecdotes, or lean on humor:
“𝘓𝘢𝘴𝘵 𝘵𝘪𝘮𝘦 𝘐 𝘤𝘩𝘦𝘤𝘬𝘦𝘥, 𝘐 𝘸𝘢𝘴𝘯'𝘵 𝘢 𝘳𝘰𝘣𝘰𝘵, 𝘣𝘶𝘵 𝘭𝘦𝘵 𝘮𝘦 𝘥𝘰𝘶𝘣𝘭𝘦-𝘤𝘩𝘦𝘤𝘬 𝘮𝘺 𝘤𝘰𝘥𝘦.”
But AI is already learning to joke, stumble, and chuckle convincingly. The line keeps blurring.
My cash problem was simple: London ATMs really do insist on plastic.
Dinner got paid through my digital wallet, no sweat.
But that split-second doubt — AI or human — stayed with me.
𝗧𝗵𝗲 𝗛𝘂𝗺𝗮𝗻 𝗖𝗼𝘀𝘁
Behind every AI-enhanced call is a person who wants to be trusted. When agents spend shifts defending their humanity, it erodes something fundamental.
We coach them to sound flawless, then penalise them for not sounding "authentic" enough.
𝗪𝗵𝗲𝗿𝗲 𝗟𝗲𝗮𝗱𝗲𝗿𝘀 𝗖𝗮𝗻 𝗦𝘁𝗲𝗽 𝗜𝗻
Leading a customer support team? Consider these:
✅ 𝗖𝗲𝗹𝗲𝗯𝗿𝗮𝘁𝗲 𝗵𝘂𝗺𝗮𝗻 𝗾𝘂𝗶𝗿𝗸𝘀 – Let agents speak naturally, even imperfectly.
✅ 𝗕𝗲 𝗿𝗮𝗱𝗶𝗰𝗮𝗹𝗹𝘆 𝘁𝗿𝗮𝗻𝘀𝗽𝗮𝗿𝗲𝗻𝘁 – Inform customers when they're talking to AI.
✅ 𝗘𝗹𝗲𝘃𝗮𝘁𝗲, 𝗱𝗼𝗻'𝘁 𝗲𝗰𝗹𝗶𝗽𝘀𝗲 – Let AI handle routine tasks so people deliver genuine empathy.
𝗧𝗵𝗲 𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆
We’re inching toward a world where proving you’re human is part of the job description.
AI should reveal our humanity, not conceal it.
Let’s remember who we’re serving and who’s doing the serving.
That's what I think. What do you think?
