"Your call is so important to us, we have fired all the humans and replaced them with a terrible automated system that cannot understand you.

Please hold while we pay our executives another bonus for some reason.

Did you know you can use the Internet to discover our website can't answer your question?"

@daedalus this was the last straw for me and BofA
Tried to change my address in the app, and when it didn't work the app gave me a number to call. The phone bot asked me what to do, I said I tried to change my address but the app wasn't working, so it told me how to change my address in the app and then immediately hung up on me 😭

@raphaelmorgan @daedalus I see they've done great things with the system I helped engineer for them. /s

I was actually proud of what I'd built, at one point.

@hosford42 @daedalus ooh what part did you work on? I can imagine this technology being used in many less frustrating ways
Like, imagine if it took a few words to try to figure out *who to send me to* and if it can't figure out what I want it just puts me on the line with a person? That seems like it would be a great use for this phone system, and it could actually save a lot of time! But no we gotta get rid of the humans as much as possible 😩

@raphaelmorgan @daedalus The NLP tech. Intent detection, entity extraction, and even a novel anaphora resolution algorithm that is patented with me as an author. (Not that I own it. I was required to give up all rights, while still being held indefinitely responsible for helping them defend the claim.)

It used to route you to an actual person if it couldn't help. But of course, they promptly made it a goal to minimize that because it's expensive.

@hosford42 @raphaelmorgan @daedalus oooo, someone who worked on these things?! I have to ask you about my cargo cult.

I tend to deal with these systems by cutting off the prompts and swearing at them very loudly. this often seems to get me routed to an agent. are they designed to detect this sort of behaviour and give up? or am I just imagining it?

@adamw @raphaelmorgan @daedalus I didn't work on this, specifically, but a lot of these systems actually have emotion detection built in, to recognize when someone is getting infuriated and redirect them before it's too late. Sometimes, when I can't figure out how to get to a person and I really need to, I will get sweary on purpose to trigger this. They are inadvertently training us to be assholes. :/
@hosford42 @raphaelmorgan @daedalus oh wow, so it actually works. :D thanks! and yeah, I know what you mean about handling the redirected calls of people who are *genuinely* angry. I once worked a call centre job which was *supposed* to be just support for install techs but somehow after a few weeks all the calls from customers mad about overage charges started getting directed to us too. that was such fun :/
@adamw @raphaelmorgan @daedalus I worked as a dev for a debt collection call center years ago. I've seen first hand the misery it causes, not just to the "customers" but to the folks forced to call them. Countless people tormented on both ends, all so some corporation can make a few bucks.