Today I Learned

I do call center work in a health care environment. We get lots of scams. Most of them are bad and obvious but someone recently did the math and figured out the don’t need to be good to work.

Follow me for a moment.

Call comes in. It’s a recording. You know this recording. It’s a busy office environment. Paper rustling, typing, annoyed sigh exactly the same number of seconds in every call.

There is no response to your voice. But you have to say the same thing 3 times with no response before you can disconnect the call. So the recoding loops and you continue talking to the bot.

Why?

Well on my side, I know it’s dumb but I have to do it because metrics mean I can continue to almost afford to do things like eat food or masturbate in a warm house in the winter.

They do it because this bot lets them map out our IVR (whatever, it sucks now that it’s AI) and capture voice samples from people who are forbidden to hang up.

Now in years past this wouldn’t be all that useful. The samples are of reps saying basically the same damn thing. But we now live in the era of lifeless AI. So the bar has been lowered for what a legit interaction is. (Seriously, some places paid extra for a more “lifelike” AI that did everything the old EVA bot did but in an Indian accent with the sound of crumpling paper in the background and the occasional “um” thrown in.)

So those voice samples can be used to create a fake call center based on real employee voices. This is a known attack vector that is being used against us in health care right NOW.

But AI needs to profitable so nothing is done about it.

Seriously, they protect AI to such a ridiculous extent they know the scam is happening from the same phone number and they won’t block it or even issue it a challenge.