How online dating is getting a Gen AI overhaul

In an Instagram ad post by the Aisle Network dating app, Shruthi, 28, smiles while an image caption recalls her “amazing” chat with Ajay on the platform as she wonders whether she is finally beginning to understand romance. In another instance, Kruthika, 25, tells people to resist peer pressure and use the platform to find a partner instead.

Shruthi’s beaming face is “highly likely to be AI generated,” per an online image detector.

An analysis of the ad post’s subject, using an AI image detector
| Photo Credit:
Screenshot taken from Aisle Network on Instagram, and isitai.com; graphics compiled on Canva

Meanwhile Kruthika’s image yielded a similar result.

An analysis of the ad post’s subject, using an AI image detector
| Photo Credit:
Screenshot taken from Aisle Network on Instagram, and isitai.com; graphics compiled on Canva

Since AI-image detectors can also yield false positives, The Hindu reached out to the dating company in order to verify the results. However, Aisle Network did not respond to multiple queries about whether the ads were actual pictures of the company’s customers, or AI-generated personas.

Dating apps actively use social networking sites to promote their platforms, but the surge in new, unregulated Gen AI tools means even small-scale businesses can use text and image generators to push their products and compete with larger companies.

(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)

In the past, dating sites used AI to match profiles and send targeted feeds to their customers. Bumble, one of the top dating apps with over 50 million downloads on the Google Play Store, confirmed to The Hindu that it uses AI for tasks ranging from flagging scams to showing users better matches.

“We have utilised AI in Bumble for many years, and we’re excited by the features we have launched to date, as well as the developments and innovations our team is focused on creating to give our community the best experience possible on the app,” Bumble said in a statement.

The dating company uses AI in its ‘For You’ algorithm to curate “the most compatible” profiles. The company also uses the AI-powered ‘Deception Detector’ along with human intervention to identify fake profiles, while a ‘Private Detector’ flags potentially lewd images.

However, with the rise of Gen AI, malicious actors have a new range of cheap, easy-to-use tools at their disposal, while dating platforms are playing catch-up in order to weed out bad actors—before they ruin lives. In 2023, Bumble worked with the non-profit ‘Partnership on AI’ to help create a framework for the regulation of synthetic media on its platform. Bumble admitted that user profiles created with AI were becoming a problem, leading to the impersonation of known people or AI-generated photos of non-existent people.

“In our automated Quality Assurance (QA) and manual moderation queues, we started noticing more cases of potentially harmful profiles seeking verification (sometimes successfully) with pictures that were later confirmed to be inauthentic and possibly generated by AI,” noted Bumble in its case submission.

Bumble also provided an example of how humanlike a deepfake could look, and how such photos could even be used by dating app members to bypass the verification process.

Examples of deepfakes provided by Bumble, in the case study ‘How Bumble is preventing malicious AI-generated dating profiles’ with the Partnership on AI organisation; note the missing thumb in the second picture
| Photo Credit:
Bumble’s Case Submission

AI as a marketing gimmick

Though Bumble has extensively reported on its use of artificial intelligence to provide a better customer experience, other dating apps use ‘AI’ as a vague buzz word to boost their advertising.

One example is the Betterhalf.ai Indian matrimony app, which claims to provide the ‘Most Compatible Matches with AI matchmaking’ and “2x Faster Matchmaking” in its Google Play Store description.

Responding to a negative user review that claimed the app’s partner preference filters did not match what was shown to the user, the company said, “Dear User, Our AI algorithm has identified potential matches for you within our community based on your profile and activity. While these recommendations may not align perfectly with your partner search filters, they do fall within compatible parameters,” before providing a phone number for assistance.

At a tech conference this year, Bumble’s founder and executive chairman Whitney Wolfe Herd floated the idea of an AI dating concierge that could assist users with dating while reducing the need to talk to actual people.

At the same time, the former CEO of Bumble further emphasised creating a platform for not just romantic users.

“We will not be a dating app in a few years,” she said during the conference. “Dating will be a component, but we will be a true human connection platform.”

However, accountability is still a looming question mark when it comes to dating and matrimony apps. Once businesses decide AI is fair game, how can users be sure they are really interacting with a human on the other end, or looking at ads featuring real people?

When asked whether Bumble used AI-generated images or videos of people in order to promote the platform or its services, the company declined to respond.

Tinder, Boo, Hinge, Aisle Network, and Betterhalf.ai were all contacted for comment about how they use AI in their dating apps, but did not respond to The Hindu’s questions.

Tech and AI safety tips for dating app users

When downloading a new dating or matrimony app, check the app’s data safety section in the app store to see how your personal information will be handled and stored

If you feel a user profile on a dating site is fake or AI-generated, report it to the app in question and do not interact with the profile or its owner

Typical signs of a Gen AI photo include: vague and excessively blurred backgrounds without any local details, merged or missing fingers and hands, a lack of identifiable clothing and jewellery details, excessively air-brushed skin and no pores, an absence of realistic shadows, strong resemblance to a celeb’s face, or a strangely cropped image that hides watermarks. Gen AI deepfakes generally have a soft or “creamy” appearance, while those taken on smartphones and cameras have graininess, varying sharpness levels, textures, flaws, uneven lighting, etc., even when edited

Check a dating/matrimony app’s social media profiles to see how much Gen AI media the company uses in its advertising content and websites. Be on the alert if their text feels too generic, if customer testimonials are not linked back to real people, or if photos look overly edited and unnatural

Never use Gen AI tools like ChatGPT or DALL-E to write your own dating app bio or create profile photos, as this is dishonest and potentially illegal

You can screenshot and upload photos to an AI image detector to see if they are likely AI-generated or human-generated. While such results are usually not 100% accurate, they may help you decide whether to interact with the account or ignore it to be safe

Do not pay for a dating/matrimony app’s premium tiers, paid versions, or annual subscriptions unless you are extremely happy with the quality of the free experience

Leave a Comment