Skip to content

When the AI answers the Phone

I’m sure you’ve been there: Called the 1-800 number of a large organization and been greeted by the faintly synthesized, almost human, voice of a machine. It wants to direct your call, but babbles relentlessly, spouting information that seems designed to make customers hang up. Most of us don’t need an AI to tell us to check the website/app.

Usually, dialing the 1-800 number comes as a last resort, after the website, reviews, talking to friends, reviewing your account and trying to find the information anywhere else. I reserve my calls for things that aren’t easily resolved with a pull down menu. The phrase “did you know…” followed by the many useful features of the website might be helpful if you’d never heard of the internet or the company before but is mostly annoying when you know you have to talk to a human. Why would a corporation want to aggravate its customers with such tedium?

If you make it past the barrage of unhelpful information, the virtual phone assistant attempts to engage you in conversation. This is as frustrating as talking to a 2 year old, but less endearing. 

There are the AIs that only pretend to understand you. 

For example, an AI asked me the nature of my question. I answered:

‘Home insurance.’

And it responded:

‘Ok, you want to make a claim.’

Despite telling it that isn’t what I want, it carried on, assigning me to the wrong queue.

There are the AIs that can’t put your response into their predefined categories.

I’ve had many an AI tell me it didn’t understand my answer to its question, then coach me on what things I should say. 

“I’m sorry, I didn’t understand that. Say things like yes, no, or renew my subscription.”

This suggests I’m the one that doesn’t know how to communicate in the language we’re speaking. One I’ve been speaking for about six decades, and the AI doesn’t completely understand because it hasn’t had enough samples to train on yet. 

There are AIs that just give up on you.

In one incident, after I’d repeated my choice several times, the AI said:

“I’m sorry, I didn’t understand that.”

Then it stopped talking. Was I to conclude I was ineligible to obtain the services of this business1?

This reminded me creepily of the famous scene in 2001 A Space Odyssey, when the computer that controls the space ship locks the human out, because the human was planning to turn the computer off. I wish I could turn the AI answering thing off. I can, effectively. I can choose to take my business elsewhere. 

The AIs I’ve talked to are bad at listening. They won’t stop talking (repeating the information they are sure you’ve missed on the very useful website or the wonderful features of the app) and listen to your answer to its questions. This is a gating issue. A machine should be able to listen and talk simultaneously. Humans at least know when to stop talking and listen when they can’t do both at once.

Oddly I’ve found that when the virtual assistant asks me to answer yes or no, pushing 1 answers yes and 2 works for no. And pushing zero often gets me transferred to the human queue. It’s like the virtual assistant’s first language is digital, which it responds to better than human vocalization. 

All this frustration begs the question of why businesses are using AI for first contact with customers.

A fundamental premise seems forgotten by organizations that delegate answering the phone to a machine: providing good, efficient customer service retains and gains customers. Really good services lets your customers sell the business to their contacts. What’s not to love about providing good customer service?

Businesses utilizing AI interfaces mean well. Efficient greeting and routing of customer queries to the right department, can serve customers faster and minimize costs. This is good business and good customer service.

Why then, have I had so many miserable experiences with the the AI that answers the phone in my quest for customer service? As a business analyst, I would suggest: bandwagoning, adopting weakly understood technology, and insufficient product testing. In hard-to-find-employee times, there is pressure to serve customers with fewer and fewer employees. AI seems like a good solution. 

I suspect AI’s phone answering time hasn’t come yet. It’s like the 13 year old who looks mature and thinks they know all about adulting, but can’t drive, doesn’t have a credit card, any debt, or an ex who wants custody of the kids. AI isn’t at the state of maturity where it can understand the breadth and depth of human communication. 

The goals of service and efficiency through use of artificial intelligence to direct incoming calls are reasonable. Turns out that’s hard to achieve. Until the truly understanding AI is engineered, pushing ‘1’ for savings, ‘2’ for chequing, and ‘3’ for other is efficient, customer centric and free of misinterpretation by a not-intelligent-enough machine. 

1 which happened to be a company I’d had an account with for over 20 years. 

Thanks for reading.

If you'd like more,

sign up to receive awesome content in your inbox, every week-ish.

Signing up is only for updates when new blog posts are added to my site. If you want marketing spam, you'll have to look elsewhere, but, really, anywhere will likely do 😎

Share this post, if you like...

Leave a Reply

Your email address will not be published. Required fields are marked *