It wasn’t too long ago that I was trying to find love in all the wrong places. I was browsing through dating websites, endlessly swiping on dating apps, and even visited an overpriced dating agency. While all of these were supposed to help me find my perfect partner, often I found myself being faced with matches that weren’t right for me.
Based on my negative experiences, I wanted to make it easier for people to find the right partner, to create a dating app that would put the users first and show them personalised, curated matches using AI.
At the time, I had a stable banking job but always wanted to be an entrepreneur and transform the dating industry. So, I set out to put my thoughts into action and started to do my own research.
I spoke with more than 100 millennials, asking them what they wanted to see in a dating app and built a dating app from the bottom up. I also did my own research to find out how the approach to matchmaking has changed over time.
The evolution of dating: From paper ads to AI algorithms
When the first newspaper was published in 1690, it gave rise to the earliest ads. Among these were personal ads by bachelors searching for eligible wives (dating as far back as 1695). Fast forward to the cuffing season of 1965, when two Harvard students used an IBM 1401 to create the first computer-based matchmaking service, Operation Match.
The idea caught the eyes of 90,000 love-hopeful singles, who received a 75 question survey through the mail and were asked to submit their completed form (along with a US$3 fee) for a list of computer-generated matches.
From there, the world’s first online dating website, Match was launched in 1995. The initial version listed online personal ads and singles could randomly search through the site’s active profiles to find a match. Talk about finding a needle in the haystack. Dating websites have evolved into dating apps driven by algorithms and artificial intelligence (AI). Algorithms are pieces of instructions, or code, that can tell an app how to accomplish a specific task.
Also Read: Finding love in the pandemic-stricken world: How online dating has changed for the better
In the context of dating apps, the algorithm starts learning more about a person from the moment they create a profile. Every decision and interaction on the dating app becomes part of a larger maze. It’s a live feedback system, where you are constantly rating and being rated
by other users based on data.
However, the algorithms behind most common dating apps find a match based on filters like age and basic interests, matches are shown at random with little to no curation.
Dating apps and data: Where does the data come from?
The first step to understanding how AI dating apps work is to look at their data pools. Algorithms process data from a variety of sources, from information that we share to how we interact with the platform. For example, many dating apps in Singapore suggest signing up using
Instagram or Facebook, feeding the algorithm initial data about a user. Many people willingly connect their social media with dating apps because it’s an easier way to sign up and share more about their personality and interests.
When comparing the dating apps in Singapore, I noticed a gap in the market; matchmaking apps were largely superficial and encouraged mindless swiping. What if there was a dating app for people to connect based on more than just a photo? What if a dating app could match people based on similar traits or life goals and even nudge people to meet offline, by suggesting nearby date ideas at a cosy cafe or a new restaurant?
That’s what motivated me to create MatchMde, an AI dating app that takes the sign up process a step further by asking personality-based questions including a person’s love language, how they describe themselves, and how they view the world. Dating apps use personality based data to show users compatible profiles using either content filtering or collaborative filtering.
Content filtering vs collaborative filtering
Content filtering provides recommendations based on user preferences. This is largely determined by individual swiping history. Collaborative filtering is when the algorithm bases its predictions on the user’s personal preferences as well as the opinion of the majority.
When you first start using a dating app, your recommendations are almost entirely dependent on collaborative filtering, or what other users think. It’s the same type of recommendation system used by Netflix or YouTube, taking your past behaviours (and the behaviour of others) into account to predict what will keep you engaged on their platform. So, everything you click and interact with on a dating app is detected, tracked, and stored as part of a constant feedback loop.
Also Read: You’ve heard about speed dating. Here’s what you need to know about speed hiring
I wanted to use technology to eliminate unfair bias by programming the algorithm in a way that encourages users to submit personality assessments, date feedback, community feedback, user behaviour, and more. When users have access to information like feedback from other users, it promotes a safer environment for everyone on the dating app and also helps people make better decisions when deciding whether or not to meet a match.
AI, machine learning, predicting matches
A newer (and more exciting) development in dating apps is their ability to recommend profile matches based on AI and machine learning. AI is the science of simulating intelligent behaviour in computers, enabling the latter to exhibit human-like behavioural traits like reasoning, common sense, and decision-making.
Machine learning is a branch of AI that enables computers to learn from information without being explicitly programmed. Machine learning usually involves classification, clustering and prediction, like predicting user behaviour.
For people using AI-powered dating apps, this means a higher chance at receiving quality matches, as opposed to endless swiping and filtering through unfavourable matches or fake profiles. Algorithms learn (and improve) based on user feedback and since personality is what keeps a couple together in the long-term, the algorithm should focus on presenting matches with complimentary personality types and similar love languages.
For example, after a user goes on a date, they’re invited to rate their date. A positive rating teaches the algorithm to show you similar profiles, while a negative rating means that the algorithm will show you other profiles with complimentary personality types. It’s a constant feedback loop, where the machine learns more about the preferences of a particular user until it is successful in finding you the right match.
For the love of AI dating apps in Singapore
Comparing the personal ads of the 1690s to the matchmaking surveys of the 1960s, dating apps are really just the latest manifestation of how people are doing what we’ve always done —creating new ways to communicate to find love and companionship.
Despite having an algorithm crunch the numbers, find patterns, and make recommendations based on our behaviour, there’s still a lot about dating and relationships that an AI algorithm can’t predict: a life-long relationship goes beyond algorithms and dating apps. True love happens offline, but AI helps us take the first step by learning our preferences to show us exactly what we like and filter out the
rest.
–
Editor’s note: e27 aims to foster thought leadership by publishing contributions from the community. Become a thought leader in the community and share your opinions or ideas and earn a byline by submitting a post.
Join our e27 Telegram group, FB community or like the e27 Facebook page
Image credit: Magnet.me on Unsplash
The post Is AI the future of dating in 2021? appeared first on e27.