🤔“Can’t an AI do that?” by Celia López
My job was already being routinely replaced by AIs long before ChatGPT hit the big time. And yet, here I am, still getting paid for something that, supposedly, a machine could do more cheaply and efficiently than me, an imperfect and expensive human.
You may be wondering why. So am I. This piece is my attempt to find the answer.
I work in online support for dashboard-building software. A client writes an email describing their issue with the product, and my job is to help them solve it. I don’t touch anything myself. I just ask questions and tell them what to do based on the information they give me.
So yes, I’m basically a flesh-and-blood ChatGPT. But I understand the words that come out of my mouth (or my keyboard): they are connected by intention, not just by probabilistic models. I also understand the product showing the issue and the person experiencing it. I understand. Full stop.
I’m not saying this out of spite, I genuinely think this is key to understanding why I still have my job. I’m a support agent, but I’m also a user of other products and services. I’ve had my phone on speaker on the table, responding to machines with numbers and monosyllables until, if I was lucky, a human voice finally came on the line.
Some of the emails I receive could probably be solved by a highly accurate, well-trained chatbot, fed with the right data and updated frequently enough. And even in that ideal scenario, my presence would still add value. In some cases, because defining a problem (if you’re not familiar with the system where it occurs) can be hard. In fact, part of my job is guiding the diagnostic process, asking the right questions to identify the issue as quickly as possible. It’s working with the client to gather enough information for ChatGPT to get the answer right. In LinkedIn terms, you could give me the title Support Prompt Engineer.
The other part of my job is making the user feel safe and important. Call me an idealist if you like. Call me Marketing.
I convey reassurance because I represent the ability to solve problems that haven’t happened before or that don’t respond well to standard solutions, in other words, data that hasn’t appeared in training. Compared to the AI model behind a chatbot, I represent greater adaptability and faster learning, because my resources aren’t limited to what’s already documented, and other information sources (also known as the humans in the Development team) are easily accessible to me.
Beyond that, I’m also proof that the company understands no product is perfect: that things will need improving and fixing, and that it is available and willing to make the necessary changes.
We could counter-argue: a chatbot could be trained to handle those exceptional cases and only send an email in those situations, right? Everyone else can manage with the documentation and the chatbot. Support team salaries get cut.
Do you remember that phone on speaker while a machine was talking? Do you remember reaching the point in the call where you start repeating “TALK TO AN A-GENT”, each time with more frustration and volume?
How do you feel in those moments? Do you feel important? No, right? Quite the opposite. You feel dismissed. To the company you’re a customer of, your problems aren’t worth even a quarter of an hour of a call-centre employee’s salary, which probably doesn’t even reach three euros. I suspect, we all suspect, that most companies make more profit from each of us than that.
We could debate whether, with advances in AI, the quality of solutions offered by a human agent and a chatbot will eventually be the same. But artificial support will never be equivalent to human support. A chatbot’s availability is perceived as unlimited, whereas the availability of a human who makes you feel heard is scarce, and therefore, more valuable.
When, as a company, you replace your entire support team with an AI service, you may be sending a message of innovation and financial performance to investors. But the message you’re sending your customers is that, from now on, they’re on their own.
Celia López - Customer Succes
Want to know a little more about who wrote this article? 👇
Celia López Monreal
She studied Psychology and completed a Master’s in Research Methodology so that the questions she never stops asking — herself and others — are as useful and accurate as possible. For a while, she applied this search for answers in Data Analytics, and she now applies it to debugging code and software.