I guess the machines are taking over the dating apps, too.

About a week and a half ago I was innocently minding my own business when I got a rather mysterious text:

I don’t necessarily think of myself as a gullible person, so I figured this was probably not true. However, my curiosity was piqued for the following reasons: 1) it knew an abbreviation of my name I don’t typically use 2) it knew my relative school year (I’m a senior) 3) it knew I was gay 4) it knew I was in the performing arts… though I suppose that’s not such a stretch given the third reason… and 5) other apps like Partiful allow you to be picked by a “crush,” whose identity is only revealed if you also pick that person as your “crush” (a feature I am certain nobody uses).

All of these seemed to more strongly point to a concerning amount of information being easily gathered about me online and/or through my phone number, which troubled me deeply. Additionally, a rather important reason not listed above is that I actually received a different unsolicited dating app text within the same ten minutes stating I’d been “invited” to their app by a current user. (I found from Googling that this app requires users to invite a certain number of contacts before they can use it, so I had been randomly selected by someone who happened to have my number, and the timing was entirely coincidental). All of these are nothing on their own, and are still kind of nothing altogether, but nonetheless classes hadn’t started yet and, yes, I was bored enough to investigate. Detective Bwog is on the case. The linked site is real and associated with Ditto, an AI-based matching site. This was our first clue. The name was familiar, but I couldn’t remember why…

I wasn’t thrilled about forking over my info and personality questionnaire data to anything powered by AI, but it was, alas, the only way to make progress in my investigation. My absolute least favorite aspect of this form was the segment where the user is soft-locked into picking the “most attractive” of four uncanny AI-generated candidates. I actually switched to my laptop at this point because, even though I kept clicking the “none of these” option, I was fat-fingering some of the choices… or so I believed. In actuality, it locks you out of “none of these” after selecting it four times in a row, something not clearly indicated on a mobile device (which the site is clearly designed for! Bad UI/UX design! Bad!!). I personally found this unsettling for a number of reasons, namely that people don’t look like AI, nearly all these AI girls were conventionally attractive femmes, and my own general objections to AI-generated imagery. Also, like what I suspect is a higher than reported number of people, I don’t really have a physical type, so this was a pointless exercise. Let it be known a major flaw of my character is my susceptibility to the sunken cost fallacy, because I continued onward despite all this.

Here’s our next clue: the questionnaire asks you for some sort of statement, hint, fun fact, or whatever about yourself. I did not like this.

It also asks you for at least 2 photos of yourself, I guess so they can match you to someone else’s AI appearance preferences? I did not like this either. Hope you enjoyed my random Instagram reel screenshots, Ditto.

Anyway, after completing this, I was prompted to send a text to a number different from the one that had texted me. It prompted me to fill out the same damn thing again, so I tried manually typing out my user code, and got a very patronizing response capped with a crying emoji asking me to submit the form. From this, I surmised I was talking to a chatbot programmed with “relatable” Gen Z slang and typing patterns, i.e. a valuable suspect for interrogation. After doing the form again, which automatically filled itself with my previous answers, I began my questioning. 

Picture me in a gray trench coat, taking a long drag of a cigarette before shining a desk light in its metaphorical eyes. Every question I asked was artfully dodged in all lowercase, with sobbing emojis peppered throughout, like footprints upon the sand. After a few unsuccessful attempts, I learned it had found me from my filling out the Yak Match many eons ago (something pushed by Sidechat to match friend groups, individual friends, and romances) which was powered by Ditto… something I did not read closely enough to realize at the time. It kept giving me the runaround, but I managed to get this out of it: Ditto used my data from the Yak Match to find someone I might be matched with based on their answers to their new questionnaire, and sent that person’s entered statement. No actual match had been made, and whoever wrote the “hint” I received had no idea it would be sent to me, nor do they probably know who I am. The fact that this person shared similar activities to myself was not known or intended by the AI.

So there you have it. An extremely minor mystery that did not warrant this level of investigation has been solved, and Detective Bwog lives to sleuth another day. In my individual opinion, these algorithmic matching things are kind of ineffective because the best case scenario is getting matched with someone unnervingly similar to yourself, but I concede some people are into that. If you don’t mind telling your kids you met their mother via Ditto AI, then more power to you. I just think Columbia Cupid’s wings are better left clipped.

Header via Author