Just as the culture seemed to collectively burn out on dating apps, AI is promising to ride to the rescue. Most established companies in the space, including Hinge, Bumble, Tinder, and Grindr, are adding AI features, including AI-powered matchmaker services and “wingmen.”
They’re joined by a host of startups offering their own versions of an AI matchmaker who can sort through the endless options on these platforms and find you the perfect partner (or at least a shortlist of contenders).
These new high-tech dating tools have an obvious appeal, offering to reduce or eliminate the soul-destroying grind and constant disappointment of traditional online dating. They’ve certainly giving beleaguered dating app companies a much needed boost.
But there’s one big problem with services and apps promising to use AI to help find you love, according to people who actually study the science of relationships for a living. Research has yet to find any reliable way to guess who you will be compatible with before you meet them. AI is very unlikely to do any better.
Will two people click? No one knows
AI is an incredible tool that’s vastly better than humans at spotting patterns in massive quantities of noisy data. But no matter how powerful, AI can’t spot patterns that don’t exist. So the question is, are there patterns that can predict who we will click with?
Amie Gordon, a social psychologist at the University of Michigan, has dedicated her academic career to studying what makes couples compatible. Which makes her about as well positioned as anyone to lay out the current state of science when it comes to predicting whether two people will fall in love.
And according to Gordon essentially no one knows much of anything concrete (yet) about why we fall for one person and not another.
“Sure, we know some of the factors that generally predict liking—physical attractiveness, positive personality traits, similarity, familiarity, reciprocity—but we have yet to predict with the kind of certainty dating apps might like to claim whether any two people will hit it off before they even meet,” she reported recently on Greater Good.
[..]
What should AI matchmakers even look for?
Just matching people up based on the basis of similarity or stated preferences is pretty clearly not effective then. But perhaps there is another way to predict who will fall in love that humans, with our limited brains, simply haven’t figured out yet. This is just the kind of problem AI is ideal for, after all. One of the technology’s greatest strengths is taking huge, seemingly chaotic piles of information and finding a faint signals in the noise.
That’s why AI can spot patterns that human doctors miss. And also why I would put more stock in an AI-driven weather forecast than an old guy predicting rain based on his creaky knees. But there still has to be some signal in the noise. Relationship researchers are divided on whether predicting romantic connection is even theoretically possible.
Some believe love will prove even more anarchic than the weather. Others, including researchers at Gordon’s lab, are trying to combine dating app data and real-life outcomes to figure out if predicting an offline spark from online profiles might be doable.
Maybe these efforts will yield results and AI matchmakers will one day be a miracle technology for the lovelorn.
In the meantime, “we must always remember that AI is trained with specific beliefs about what makes two people compatible,” Gordon reminds readers. At this point in time no one — not even leading researchers in the field — have the foggiest idea what those beliefs should be.
