In the second decade of the 21st century, we can observe the burst of AI consulting services and startups in versatile industries. Machine learning algorithms have already made a good showing in recommender systems, data security, computer-assisted diagnosis, self-driving vehicles, and other miscellaneous use cases. And a regular user might not be even aware that this subfield of artificial intelligence has been actively leveraged to end up with his or her singlehood. So, next time when you swipe someone in Tinder, be sure that progressive technologies do care about the matters of your heart.
Why online dating?
If you suppose that implementing such a stunning technology in online dating is a lack of common sense, you might have lost track of modern trends. Today, dating app users are not only seekers of love or soulmate. People use random video chatting websites to find friends, romances, and sexual affairs. Hence, communication is what attracts millions of people to install and use a dating app. According to Statista, almost 20% of the US Internet users are into online dating. If we look globally, this share is smaller – almost 4% of all people with access to the Internet which is around 290 million users worldwide, use dating apps! Moreover, along with a clear-cut age group of users (18 to 30 years), older generations also use their smartphones to communicate with others or search for a soulmate. In other words, this niche is a rather lucrative business which attracts investments and trendsetting technologies hereinto. Probably, that is why the paths of artificial intelligence and online dating have crossed.
Top popular dating apps powered by machine learning
In 2012, the number of websites for dating amounted to 7500 globally. The trend kept evolving and went over to mobile applications. There is a plenty of them on both App Store and Google Play. And what dating apps are powered by machine learning? Here is our pick.
Even those who have never used Tinder to strike matches heard about its matching principle – swipe right if you like someone and left if you don’t. Users have several photos in their profiles, and the dating app can detect the most attractive one by itself. Smart Photos, an ML-based feature, learns how many times each photo was swiped right and makes the corresponding rating. As a result, the photo with the highest swiped-right rate becomes your profile picture. Another algorithm analyzes your typical swiping preferences (what types of photos you like) according to different metrics (geolocation, overall swiped-right rating, etc.) and suggests those you’re likely to swipe right as well.
Hinge has recently joined the cohort of dating apps that use machine learning. A new feature called The Most Compatible relies on a matching algorithm developed in the year when the first American astronaut orbited the Earth. The ML-based system analyzes users behavior on the app (liked content, users, photos) and suggests matches with users having similar behavior.
The official counter of Badoo users indicates around 400 million people so far. We pay attention to this service due to its facial recognition technology powered by AI. For example, you want to meet a girl with facial features similar to Scarlett Johansson, the system will make a selection and suggest users with the highest similarity. Be careful because what is shown in photos and in reality are not always equal:)
Many users go with Tinder like a game due to its swiping feature. For this reason, another category of users opts for Dine, which is a more result-oriented solution. In our case, the result is a date. To achieve it, you need to attract another user with your best photo which is not always the one you like the most. Don’t worry. A smart AI-based system will select a photo that attracts others most, and offer to make it your profile picture. Experience shows almost a dual growth of matches if you follow the algorithm’s recommendation.
AI-based photo rating at OkCupid is akin to Dine. However, this dating service has another smart algorithm to help users find their matches. Apart from uploading your visual data (photos), you have to answer a number of questions. These answers shape a specific user profile, which is then verified for compatibility with other users. In other words, the more things two users agree on, the higher the probability of a match. That’s why the number and accuracy of answers are essential to find the best match for you.
What does machine learning allow dating apps to do?
Most of the online dating apps use similar machine learning algorithms that analyze tons of user-related data including personal preferences, location, likes/dislikes, clicks, photo rating and others to filter out the most compatible matches. What else can machine learning offer to online dating?
Find better matches
Just like a recommender system at Netflix or YouTube predicts videos you might be interested in, ML algorithms in dating apps analyze your activity to find better matches. However, it is more complex here since the analysis is not limited to one user’s interests. Dating always foresees two users which, in turn, entails the necessity to find as many things in common as possible to achieve the perfect match.
Your best picture
The abovementioned Tinder, Dine, and OkCupid take care about your success in a search for a soulmate from the very beginning – your profile picture. It is the first thing that creates an impression for online daters. If one of your photos has a better attraction rate than others, machine learning will set it as a profile pic.
Fake users and time-wasters make a widespread issue of online dating as a whole. Machine learning systems weed out users that are not real by using different techniques akin to fraud detection algorithms used in online banking. For example, the Belong dating app checks users’ lexis and photos for fakeness (fake identities use primitive specific phrases and inauthentic pictures).
Locking out offenders
Unfortunately, not all users opt for online dating to find love or soulmate. There is a category of people who come here to mock, humiliate, and perform other offenses. In this case, filtering algorithms are especially valuable. They lock out offenders that use blacklist vocabulary (insulting, profanity, swear, and curse words) and inappropriate content.
We mentioned that Badoo allows you to find matches according to your visual preferences. A smart algorithm analyzes the photo you upload, be it a celebrity or unknown person whose picture you like, and recommends matches that are visually similar to your photo sample. You might fail to date Tom Hardy or Penelope Cruz, but you can find a person who looks similar to your favorite actor/actress and invite for him or her for dinner!
Machine learning does not guarantee that you find your love through a dating app. But you should not reject to give a try and discover the capabilities this technology has brought to online dating.