If you want to know what the ninth circle of Hell is like, join a dating site. I didn’t want to join a dating site, for I am happily ensconced, but I joined one in the interests of research because something very dark is going on in the world of dating apps, and it has something to do with two bugbears of mine: safeguarding and AI.
If you could mate two monsters of modern life to produce a deranged creature capable of devouring human rights, then those two monsters would be safeguarding and AI.
Recently, I began to notice a deluge of complaints on online forums from people protesting their innocence after being banned from dating apps for no apparent reason. Reddit is awash with these demented souls, some of whom are so appalled at being summarily accused of unknown offences of a predatory nature that they are presenting as nigh-on suicidal.
Girding my loins, and all that, I decided to join the app users were claiming was the worst offender, Hinge, the largest dating app in America, with 30 million users worldwide. It is owned by Match, which also owns Tinder. Match Group proudly states that in the interests of safety, 80 per cent of harmful accounts are now removed using AI before a real user ever reports them.
Now, you have to realise that in joining Hinge, I obviously set out to behave as mildly as I could. I wanted to perform an experiment, so I resolved to start out as innocently as possible, before feeling my way to flirting and maybe seeing what passes for offensive, which I assumed wouldn’t be much.
I made my profile polite and correct, uploaded professional head shots and did exactly as the app directed by ticking the ‘like’ icon on certain profiles to show the sort of person I was looking for. I made one comment on one user’s profile, which was the mildest little flirt I could think of. ‘Nice tux!’ I tried to write on a picture of a chap dressed smartly at a wedding. But the system instantly flashed up a warning that the comment was inappropriate. So I deleted it and didn’t post it. I didn’t contact or message another user. I simply viewed profiles for 15 minutes, and that was that.
The next morning, I was amazed to receive an obnoxious email from Hinge entitled ‘Your account has been removed’: ‘Hi Melissa, Your account was banned for breaking one or more of our policies. You can log in to Hinge to learn more about why your account was banned and how to appeal, if that information is available to you.’
When I logged in, it referred me only to the acres-long general terms and conditions. So the information was not available to me. I felt like Joseph K in The Trial, accused of a crime I knew nothing about, which my accusers would not detail. Actually, it was worse. Even Joseph K was put on trial. Melissa K would not get that chance. How could I decide to go through the appeals process, not knowing the offence?
I deleted the app. Ironically, the Hinge slogan is ‘Hinge – designed to be deleted’. Now I knew why. I was so paranoid, I feared that if I left Hinge on my phone it might start spreading rumours about me to my other apps. My bank might remove me as an undesirable. Well, I don’t know how these things work, do I? You can’t be too careful.
I then contacted Hinge head office to say I was a journalist, and to ask what I had done. It is at this point companies usually admit their mistake, apologise, and offer you free stuff. But after two days, a Hinge spokesman came back to me with an even more obnoxious email, rambling and bureaucratic:
‘Thank you for reaching out. While we understand you are raising concerns about your own account, due to privacy requirements, we cannot comment on the specifics of an individual enforcement in the media.’
I took that to infer that what I had done might be so bad they couldn’t possibly embarrass me by spelling it out. ‘We recommend submitting an appeal,’ he said, while a lengthy briefing with bullet points explained that every appeal is reviewed by a human moderator in a ‘Trust and Safety Team’.
These moderators have the ability either to take action on a report or ‘escalate the case’ to a senior moderator or safety specialist. I decided this could get a lot worse if I chose to appeal. I was better off remaining frozen, as in the frozen lake of Dante’s ninth circle of Hell.
Look, Hinge, the bottom line is, there is no offence. I’ve done nothing. I know it and you know it. The AI has simply goofed up. Maybe it read my tux comment wrongly or my profile hobby of ‘riding’ as an attempt at innuendo. Shows you AI is as thick as mince.
I sent an email saying I’m not appealing on principle unless you tell me what I’m appealing. ‘We understand this is frustrating,’ came back the reply, ‘and we’re sorry that this hasn’t been the experience you were hoping for.’ But it has, really, because I set out to prove the poor demented souls on Reddit aren’t lying. Hinge is moving against lots of its users. Why, I still have no idea.
It is possible they were weeding me out for reasons they cannot admit. I didn’t pay a subscription and was using the free service, so my first suspicion was that dating apps might be removing people who look as if they’re not going to cough up some money.
The second paranoid delusion that beset me was that Hinge had filtered me for demographic reasons. When I signed up, I noticed some very woke promotional shots of all sorts of people not like me finding love. I had to answer a preposterously long list of sexual- and gender-orientation questions, as well as describe my colour, race and religion.
I felt like Joseph K in The Trial, accused of a crime I knew nothing about
Thinking back on this, and feeling utterly persecuted, I put my deranged and paranoid theory to Hinge that I suspected it had programmed its AI to filter out certain categories of people in certain numbers at certain times, in order to keep its brand woke. Was a 54-year-old straight, binary woman not wanted in the mix this month? Its response was not to answer that question at all.
So let’s give them the benefit of the doubt and say this is a barmy idea they’ve driven me to imagine with their mysterious secrecy.
I would still argue that by refusing to spell out why they don’t want me, Hinge is happy to allow aspersions to remain, while challenging me to a kind of techno-duel in which I have to weigh up how lucky I feel that the system won’t misconstrue my non-offence even more if I try to dispute it formally.
This is par for the course with safeguarding. Throw some muck randomly and brazen it out is too often a company’s policy. With AI it’s even worse. Hide behind unknown algorithms to be as rude and as dis-inclusive as you fancy.
AI is the only legal way a company could tell its customers to ‘Naff off, we don’t want your kind!’. That is what AI is so good at. Being damnably rude and unfair and obnoxious and, of course, just plain old wrong.
Comments