The Washington Post recently reported what sounds like a match made in Hades: online dating sites are now hooking up with artificial-intelligence startups that generate fake faces for use in ads. The dating sites hope this will enable them to project more diversity, while the AI startups are counting on dating sites to help normalize their disconcerting technology.
Having AI-generated people lure us to dating sites might sound like one further slide into sci-fi dystopia.
But it’s not that different from what’s already going on. Dating sites have been using stock photos, models and actors in their ads since inception, as have some users. And that’s better for customer privacy than the genuine — but very creepy — strategy of borrowing users’ photos for use in ads.
Of course, using AI to generate fake people is much easier to scale. Whereas before sites had to find people who looked like their target demographics, now hundreds of thousands of such people are just clicks away at low-cost because nonexistent people don’t charge a photo-shoot fee.
There is a boundary that’s being tested here: Dating sites have gotten in trouble before for producing ads that seemed like messages from real people; there might be similar concerns if they start using fake people to populate their actual dating pools. Plus, it’s likely that expanding the prevalence of AI-generated people will eventually enable all sorts of new scams.
But if dating sites use fake people to make their dating pools look more diverse, that might actually improve the sites’ diversity in the long run because people are more likely to join platforms with others that look like themselves. As social engineering goes, that doesn’t sound so bad.
Nevertheless, today’s online dating markets face a broader problem of inauthenticity: In the era of a gamified dating, profiles often don’t contain much information, and what little data users provide is carefully curated to maximize hit rates. And while people have been lying about their height, weight, age and professional accomplishments on dating sites for decades, nowadays they’re even hiring professional assistants to manage their chats.
This means that when you meet someone through a dating site, it can be like a game of date roulette: Will the person look or sound anything like the individual you interacted with online? “Maybe” isn’t a very satisfactory answer.
And when meeting people is a gamble, that’s an indication the market isn’t very efficient: It can take a long time and a lot of work to find somebody whom you actually want to spend time with, never mind something more serious. The result is that the whole process can be pretty discouraging and feel like a poor use of time.
That really is dystopian and it has left many users — especially millennials — looking for a more authentic dating experience. It’s hard to imagine that adding legions of fake people into the mix is the remedy. So dating sites may find that AI-generated individuals just stoke user distrust. That’s never a great basis for a relationship of any kind, whether between potential partners or between businesses and the consumers they’re trying to win over.
Kominers is the MBA Class of 1960 Associate Professor of Business Administration at Harvard Business School, and a faculty affiliate of the Harvard Department of Economics. Previously, he was a junior fellow at the Harvard Society of Fellows and the inaugural research scholar at the Becker Friedman Institute for Research in Economics at the University of Chicago.