"Hey, do you like the ninja turtles?"

This was the first question I asked a girl I fell in love with when I was 16. To this day, I'm not sure how or why the armoured reptiles were what came to mind. But it was this question that set the tone for the entire conversation thereafter: confused, incoherent, and ending with me desperately saying, "Let's talk about the Ninja Turtles again sometime".

That was the last time I talked to her about turtles. Actually, that was the last time I spoke to her, period.

It was clear that I needed some help when it came to socializing with the ladies. I turned to my friends for advice, but they knew even less than I did about what girls wanted. I searched in the library but also didn't find sufficient help there.

What I am realizing now is that my son will not have the same problems I had, because he will have the help of artificial intelligence. In fact, that sort of help already exists.

Artificial intelligence is starting to reshape the dating world just as it is in other fields. Apps like Tinder and Bumble are still the most common, but you can see the next generation of ‘love’ apps emerging: one where the app doesn't just recommend matches but helps the user throughout the entire process.

The AIMM app for example provides a good example of the dating of the future. The user can’t just download it and immediately start looking for love. The app first needs to learn about the user which it does through voice conversations that span over a week. During that week, the app poses hypothetical challenges to him, interrogates him about his preferred place of residence in the future, asks what his desires out of life are, inquiries about his sexual preferences, and many other things that even the most persistent matchmaking aunt would not ask. And no, the Ninja Turtles are not mentioned at any point.

After this introductory phase with artificial intelligence, the application begins to match the users who are better suited for one another based on the data, which specified their desires and needs. The app then provides instructions on how to proceed.

Another app, Mei, also provides heartfelt advice. As the app's developers write: "Mei uses algorithms to analyze text messages on your phone." Mei can also help you interpret romantic situations, or at least those with the potential to become romantic.

Now, it appears that these two applications were ahead of their time and perhaps even exaggerated in their capabilities. Nevertheless, they provide us with a glimpse into the future: a future where even the shy and less social among us will be able to find love.

Will humans fall in love with chatbots in the future?

Recently, the dismissal of the engineer Blake Lemoine from Google made headlines, after he claimed that the company's chatbot "developed feelings" and even aspired to be recognized as an employee and demanded to be treated as a person and not as Google’s property. Google finally decided to terminate his employment because he leaked internal materials to external parties. But following the case that caused an uproar in the world of technology, the topic of human-bots interactions came up in discussion. In recent years, bot technologies have become more sophisticated to the point that they have developed unique characteristics that can easily identify the person standing in front of them, and easily provide answers and other essential information to technology companies.

But why stop there?

In a world where artificial intelligence can provide advice about love and relationships, why wouldn't companies create chatbots and virtual avatars that would be especially easy for humans to fall in love with?

This may sound like a ridiculous development: will humans really fall in love with robots and artificial intelligence? But our current reality provides hints that such a future is already on its way. Luka, for example, is a company that produces chatbots called "replicas” that are personalized for each user and accompany them throughout their lives. Some users expressed that they experience feelings of loneliness if they do not talk and share their feelings and thoughts with the chatbot at least once a day. The company's management frequently receives messages from users convinced that the artificial intelligence they are talking to is self-aware and perhaps also developed genuine feelings for them.

When the day comes that such chatbots are incorporated with a robotic body, I do not doubt that there will be people who choose to have full romantic relationships with these robots. And if not in the physical world, then most definitely in virtual reality. We hear about such cases from time to time in countries like Japan and South Korea.

Is our future looking a bit strange? Yes, but it will also be overflowing with love both between humans and robots.

The swipe game

When you open your Tinder, sophisticated algorithms start doing calculations for you. Their purpose is twofold: on the one hand, they want to serve you by connecting you with other users who are looking for love and whose chances of being loved by you are particularly high. On the other hand, they also have to serve Tinder itself and make sure that even if you are one of the most desirable 'players' on Tinder, your profile will still be displayed in a limited way, so that you don't ‘deplete the inventory’ of the other available profiles.

How exactly do these algorithms work? The simple answer is that we don't know for sure since companies like Tinder keep them top secret. From what we do know, in the past, Tinder used the Elo rating, which is a method for calculating the relative skill levels of players in zero-sum games such as chess. In general, we can say that every time someone swiped right on your picture (that is, confirmed their attraction to you), it led you to ‘win’ the game and raised your desirability score in the system.

The algorithm has undergone many changes since then, and today – at least according to the limited information released by company executives at conferences – it is called TinVec. The algorithm still learns from the swipes of each user but also considers the characteristics of the swipers and those swiped. It arranges the users in a multidimensional vector space that includes reference to the data and preferences of each of them and tries to find the closest neighbours for each user. The system is also supported by Word2Vec, the user model that uses associations and keywords from long text to learn about the user. The algorithm arranges the users again in a vector space, but this time according to the language, style and words that are repeated in different profiles.

Judging by the considerable success of Tinder we can gather that their algorithms work well in matching complete strangers. And yet, it's hard not to think about the possible negative consequences as well. Research from OkCupid showed in 2014 that in the United States there is discrimination against black women and Asian men. One can guess that Tinder's artificial intelligence – which learns about user preferences and reproduces and amplifies them – will only strengthen such negative patterns. If you are a black woman in a world that does not like black women, then your ranking will be so low that the most desirable men will never see you. If, for example, there is a stigma against wearing glasses in society, then such users will lose hope of being connected with attractive potential spouses. Artificial intelligence will actually reinforce and perpetuate existing social preferences and it will do so with zero transparency. In recent years, a lot of criticism has been directed towards dating apps because they intensify discrimination on racial or social grounds.

Is this the matchmaker we want?

When my child grows up, he will be able to find love more easily thanks to the artificial 'matchmaking aunties' that are currently on the market. They will help and guide him, but they can also narrow his pool and lead him in harmful directions like those that perpetuate inequality and discrimination. However, we must remember that many couples have and will continue to find love through various dating apps.

I'd like to think that I could trust my child to make the right choices in life, but let's face it: Humans don't have a good track record with these kinds of challenges, especially when they involve love and sex. We need to invest in educating our children to take a critical view of artificial intelligence recommendations in every field – including love – but also work to unlock the algorithmic secrets that many companies hide in their server cabinets. You don't have to reveal commercial secrets, but you do have to make sure that the algorithms don't cause damage to an individual or a company.

Written by Dr. Roey Tzezana, Researcher at Tel Aviv University and Head of Methodologies at SparkBeyond