5 things I learned by working as an automatic speech recognition transcriber

Have you ever wondered what it would be like to be the size of a ladybug and be able to randomly spy on people’s lives? Working as an automatic speech recognition transcriber is as close I got to such an experience. Except that the window I had into people’s lives was no more than 15 seconds long, and I did not have to move out of my chair to actually fly into their homes. But close enough.

I am bilingual in French and English, and moved to the U.S. a year ago. I needed an easy and flexible job while still transitioning lives and finding my feet here. I saw an ad to become a transcriber for an automatic speech recognition device, passed several tests, and started working.

I thought this would be a pretty inoffensive job, but I did not expect some of the things I would come across. Some of them pretty boring, some of them downright shocking, and some others saddening. Here’s a glimpse into the 5 main issues one may face when doing that kind of work.

1) Creepy porn requests.

Porn requests are not really shocking of a finding, are they? Still, it is a little disturbing to hear things people don’t really expect to be heard. And what is more disturbing is when you hear requests for rape scenarios or borderline child porn. That can stick with you for a few hours after hearing it, believe me.

2) Loneliness in the 21st century

Men requesting escorts or cam girls are not that weird — that kind of practice has existed since the dawn of time. Yet, it goes a step further when people directly ask to “chat” with the speech recognition device itself. One audio will stick with me forever: it started with TV in the background, and then a weak, half-asleep voice addressed his device and said “Hey, let’s chat.” The lonely tone of his voice made me shiver — or at least, what I perceived to be loneliness in his voice. Like when I hear about some people being in love with their sex dolls and things like that: when people have to resort to non-living things imitating living things in order to feel like they have a connection to the world. Indeed, the movie “Her” depicting how a lonely writer develops a close relationship with his talking operating system is not too far-fetched.

3) Sexism in the industry of speech-recognition devices…

Between Apple’s Siri and Microsoft’s Cortana, the two most popular ones (but many other “female” ones exist), how come both have a female-sounding name? Why do these “assistants”, made to serve the device’s owner, still mainly have female voices? Whether it is intentional or just so ingrained in the mentality –women are good at assisting, right– the outcome is just wrong. To go further, here is a good article in The Atlantic explaining how it is indeed based on sexist biases - see also this Time article going in more detail on that, and on the Cortana controversy. Indeed, if you want to learn a bit more about where she comes from, she is originally a character from the video game Halo, where she supports the Master Chief directly - and of course, in order to provide "tactical advice, instant-access information, and calculations that far exceed the capabilities of even a super-human Spartan soldier," she had to be naked to be able to do her job duties, right?

4) … and how that sexism is directly linked to everyday sexism

Indeed, take a female voice who “does your commands” and a dumb teenager or disgruntled adult man, and you will get many requests like “Hey, suck my d**k”, “Do you want to sleep with me?”, or simply “Shut up you stupid b***h”, and so on. And while it is “only an electronic device”, this kind of comments still says loads on how women are perceived in our society, in my view. Having more male-sounding talking devices could be a little help in shifting the general mindset.

5) The kind of “privacy” we have

Yes, I have heard several creepy pornographic requests, some of which made me sick deep in my guts. Yet one cannot help but wonder how come such private commands, which were clearly not made to be heard by another human being, came to fall into my ears. Either the device’s owner does not care, in which case that’s fine; either they actually don’t know, or realize, that their audio commands will be kept in a database and heard by another human being like me, who will need to transcribe these things word for word.

Another important point is that, with the company I work with, a button exists to report content — on grounds of criminal content, for instance. Here again, it is disturbing to know that somehow, the audio can still be traced back to the physical person in question, even though it is supposed to be anonymous. But don’t you worry, Big Brother is watching you.

This is a slightly edited version of the article that was first posted on the Medium channel "Becoming Human," which explores artificial intelligence and what it means to be human.