Ha! The Discord GDPR/Data Export thing reveals that it's running models to figure out what gender you are. If you go to /activity/analytics/events-*.json and grep for predicted_gender you get something like:
{
"user_id": "282657081457115136",
"predicted_gender": "male",
"probability": 0.8413839340209961,
"prob_male": 0.8413839340209961,
"prob_female": 0.11650349199771881,
"prob_non_binary_gender_expansive": 0.04211260750889778,
"model_version": "2024-05-08T00:00:00.000000Z",
"day_pt": "2024-05-15 00:00:00 UTC"
}</pre>Anyway, they seem to have this datapoint _over time_! Meaning you can make a graph of how male/female/NB you are according to discord, here is mine:
@benjojo I tried requesting my data a couple days ago and when I got it back today, it unfortunately did not have my Assigned Gender At Discord :(
@benjojo oooh I would love to do something like that. just requested my data, do you have something like a oneliner to generate that graph? :D
So many people will discover that according to AI, they are trans
@benjojo good find! Quite silly to store these values at all, they could just compute them as needed. Of course, why on earth are they needed?
@benjojo My export doesn't have it, maybe because I have "Use data to improve Discord" and "Use data to personalise my Discord experience" off?
@benjojo That bar between March 1st and the 15th 👀 It really seemed to be suspicious of your message contents
@wrmsr I'd love to know what did it, but I can't see anything in particular
@benjojo Is the script to generate this visualization available somewhere? I would appreciate it
@benjojo I have to say it is the most lib thing to use ai to predict a persons gender without their consent.. and then just slap a non-binary option at the bottom.
@benjojo "Assigned Male at Discord" is not something I ever thought I'd imagine or even see myself typing but here we fuckin are.
@benjojo It occurs to me that this probably constitutes medical data (because transition) and so falls into the Extra Spicy category of the GDPR
@benjojo Huh. I just got my data packet and it doesn't have a `/activity/analytics/` directory. It has the parent `activity` and under that there *are* two subdirectories (reporting and tns) each with a json file, but grepping for gender yields nothing.
I wonder if they are hiding this from users since it got out now ...
@ColetteDiskette no you probably have disabled some of the data collection
@benjojo suprised it took like 20 days for my discord export to arrive. But it just did. So here's my Discord Gender-Graph ™️
For anyone who wants to generate graphs like that (I have a line graph instead, but yeah): https://paste.lol/thibaultmol/discord-gender-graph
I can't explain what happened in March 2023 other than I went to Germany. Does Germany turn people fem?
@thibaultmol mine has a fem spike around that time too, I suspect the model was just busted
@benjojo
Funny comment from a friend who made their graph as well:
> I should put AFAD in my bio lmao
> (Assigned female at Discord)
It also seems that the same archive has a guess at how old you are too, Discord has gotten this entirely wrong, except one time.
<pre>{
"user_id": "282657081457115136",
"predicted_age": "35+",
"probability": 0.7547529339790344,
"prob_13_17": 0.0005852651665918529,
"prob_18_24": 0.014580278657376766,
"prob_25_34": 0.23008151352405548,
"prob_35_over": 0.7547529339790344,
"model_version": "2024-03-20T00:00:00.000000Z",
"day_pt": "2024-03-27 00:00:00 UTC"
}</pre>
@benjojo hmm, this feels like very dangerous territory to step on. This might either open them up to inquiries like "how dare you show NSFW content to users you predict as being 13-17", or cause many misclassified adult users to be unable to see such content.
@benjojo i have all the personalisation and excess data collection turned off so i wonder if it still does this to me. have downloaded, will see!