Student Login

Modern voice mail hell

I'm starting to think AI has made customer service worse.

When I am a customer of some company - a bank for my credit card, an airline I'm flying on, or whatever - I'll usually do what I need to do online. Because it's faster that way.

The problem is when I need to do something unusual.

Since the unusual options aren't supported through their online help, I need to talk to a human. So I call their phone number, hoping to find one.

You know about "voice mail hell", right? Where you call in, and keep getting bounced around through different menus, wasting minutes of your life not getting what you want.

Used to be, you'd just press numbers on the phone for options, bouncing around like that. But today, the voice prompt will often ask you to say what you want. And use voice recognition to decipher what you say.

Except... if I'm going to the trouble to call in, it's probably because I need to do something very non-standard. Which means it's likely whatever I ask for won't be recognized by the system.

So what happens instead?

It depends on the specific voice-mail system, but what many do is mis-classify what I say. They'll take my round peg, look at the square holes it knows, and pick one.

And it always picks the wrong hole. Grrrrr.

It's worse than the old number-punching system. At least with those, I could just not type a number I know would go in the wrong direction. But the mis-classifier shoves me down some road I don't want, all the time.

This all makes me think of UX in software. User experience.

Are we designing our programs to clearly indicate the choices to the user? And having the right fail-safes if our program doesn't support what the user wants?

It goes to the level of code, too. UX applies to the interfaces you design in your code: the classes, the methods... what they're named, the arguments they all take.

The modern voice-mail hell is a software fail.

It's applying machine learning, and we're all going to be doing that more as time marches on, not less. And for many ML tasks, Python's often a first choice of language. So as a Python developer, you're even more likely to be using it.

(In fact, that voice-mail menu code could well be implemented in Python.)

I'm thinking very long term where our profession is going. As the technologies we use evolve, of course we'll have new challenges. And we want to manage them well.

Just something to think about.

(Want to know my hack for getting past modern voice-mail hell? Just say the word "representative" in response to every single question you're asked, until you get a human.)

For Teams Bootcamp