If you’ve had any experience with dating apps, you probably have strong feelings about them. You might get burnout from spending too much time swiping, or you might meet the love of your life. And thanks to new developments in AI, you might love or hate dating apps even more.
Dating apps are no strangers to artificial intelligence. But given online dating companies’ years of experience in AI tech, it’s a welcome surprise that they don’t seem inclined to immediately flood their apps with new, ChatGPT-like features, unlike other social platforms.
From talking fridges to iPhones, our experts are here to help make the world a little less complicated.
Nowadays, AI-powered algorithms drive every major dating app. But that wasn’t always how online dating worked. Originally, online dating was targeted primarily to older audiences and required them to fill out long questionnaires about their interests, hobbies and preferences.
Then came Tinder.
“Swiping changed the online dating game … a little over a decade ago, and it was transformative,” said Liesel Sharabi, an associate professor at Arizona State University and online dating researcher.
Swiping models use foundational AI technology from fields like machine learning to develop personalized recommendations. These exist in many places online, powering things like how Netflix recommends a new show or how TikTok serves videos.
When people use dating apps, every action they make is a data point for the algorithms to determine what content to serve next. When people fill out their preferences and actively swipe, it helps the algorithm learn what content they’re most likely to engage with and what profiles make the best potential matches.
“When you swipe right on somebody to indicate that you like them, it’s going to take into account other people who have also been swiping right on the same profiles, and it’s going to use that information to then make recommendations. So it’s recommending partners based on people like you who seem to have similar tastes,” said Sharabi in a Zoom interview.
Many dating apps also use non-generative AI tools to bolster their security. In 2019, Bumble rolled out an AI-powered image detector that automatically recognized and blurred nonconsensual nudes sent in chats. In 2021, Tinder introduced a tool that scans people’s DM drafts to identify inappropriate language and sends them a pop-up message asking them to reconsider sending. More recently, Hinge created an anti-ghosting feature that prompts you to reply to existing matches, in an attempt to help mitigate burnout.
Now, the next evolution of AI technology is here in the form of generative AI. It’s the kind of technology that powers popular chatbots like ChatGPT and can create text, images and audio from simple queries. It also presents a lot of opportunities for dating apps to improve their models. Sharabi said that dating app users are in a “burnout era.”
“People are getting burnt out. They’re tired of the model that exists currently. I think we’re at a point where people are ready for some changes, and fortunately for the platforms, I think AI is going to bring that,” Sharabi said.
Leveling up your rizz with generative AI
At a tech event hosted by Bloomberg this past spring, Bumble founder Whitney Wolfe Herd said she saw a future where AI personas acted like concierges to swipe, match and message with other people’s AIs for us. That future is still a ways off in terms of technological development and implementation, but generative AI for dating apps is here.
So far, AI features on dating apps are falling into two main buckets: tools to bolster security and tools to improve engagement. Tinder recently launched an AI-powered photo selector tool, which scans your camera roll for the best pictures to include on your profile.
In the other bucket, Bumble recently announced a new AI model that it says strengthens its ability to weed out fake, scam or spam accounts. Bumble’s deception detector tool, powered by the AI model, helps people report fake profiles using AI-generated photos and videos.
Generative AI has also opened the doors for AI dating startups like Rizz Decode. Rizz is an AI-powered dating assistant that uses ChatGPT and other models to generate text for dating apps. You can upload screenshots from dating apps to the Rizz mobile app, and it will suggest replies based on what’s being discussed in the DM or on someone’s profile. The app, which is available for $10 a week, currently has over 7 million users and has generated over 75 million responses.
These are the worst dating apps in terms of privacy according to the Privacy Not Included project.
Screenshot by Katelyn Chedraoui/CNET
Grindr is one example of an app that could be affected by AI. CEO George Arison has been vocal about the possibilities AI presents, and Grindr said in a blog post it was exploring new ways to use generative AI, without giving any specifics. Arison said in an interview with Yahoo Finance earlier this year that one way AI could change how matches are made is by potentially scanning people’s DMs for additional data to improve match recommendations. Grindr has not implemented this tool, but it has updated its privacy policy to say it’s transitioning the storage of messaging info to Grindr servers, no longer just on people’s devices.
If you use Grindr, you now have to trust it with personal information, Caltrider said, and the company doesn’t have the best track record with that. Right now in the UK, Grindr is facing a big lawsuit from people who allege that Grindr shared their sensitive health information, including HIV status, without their consent. In a case from 2021, a Catholic media company bought location-tracking data from dating apps like Grindr, which reportedly led to the outing of a priest.
“It’s not only what you can read and deduce from [dating apps] policy privacy documentation. It’s also what they then turn around and actually do,” said Caltrider. “Trusting them with your information is trust that I feel a lot of these dating app companies have not earned.”
Adding AI to the mix could make it even harder to trust dating apps with our information. The rise of AI chatbots and generators has brought up many privacy and even copyright concerns, particularly around how models are trained. Big players like OpenAI have struck deals with news organizations and platforms like Reddit to use their content for training purposes. However, dating apps have been more hush-hush about their processes — including whether user data is used.
Khaves said that all screenshots on Rizz are encrypted and added that part of being on dating apps, and even other social platforms like Facebook, means forfeiting some degree of privacy. For example, Tinder simply won’t work if people don’t share their location. But when it comes to potential new AI tools, it’s going to be harder to decide to hand over data if you don’t have information about how it’s actually used — if you get the choice at all.
Generative AI on dating apps also raises new, thorny ethical questions. For example, if a dating app did start scanning DMs for data, is that something its users consent to and control? And if AI tools include human review as part of the maintenance process, are people aware of and agreeing to have their DMs potentially read by someone other than their match? Ethical questions like these, Caltrider said, are ones dating apps might not be considering as they rush to get into the AI game.
As dating apps dive deeper into AI, all of these unknowns will affect the quality of the user experience. AI promises opportunities for dating apps to improve their services and security. But it’s also likely to further cement a key issue — forcing people to give up personal information and forfeit privacy in the name of finding love.
“It becomes really tricky to implement AI successfully in a situation where every bit of personal information about you is on the table,” said Caltrider.
Future of AI in online dating
So far, new AI dating tools seem to be focused on reversing negative trends and improving the experience for daters. But there’s still a worrisome lack of information about how these models will be used and whether user data is involved in creating and training them.
There’s also the distinct possibility that all these AI tools eventually end up paywalled, accessible only to premium subscribers. A report from Morgan Stanley highlighted that monetization of people who are already using the apps — getting them to upgrade and pay for certain perks like standing out, unlocking more matches and sending unlimited likes — is likely going to be the industry’s strategy for the next few years. If that is the eventual end goal for Match Group and Bumble, they’re going to have to work to prove these AI features will be worth it.
On the other end of the spectrum, Sharabi wonders if AI might push people back to meeting more folks in real life. With AI tools becoming more common, it could be harder for people to trust they’re safe and not being catfished. AI should augment, not replace, human decision-making, she said.
“Dating apps, they’re fun and kind of frivolous, but at the same time, they’re also facilitating some of the most important decisions that you will ever make. People are using them to choose life partners. It’s a huge decision. It’s going to impact so many different aspects of your life, and we’re letting AI help us make those choices. It’s really significant,” Sharabi said.
