Can AI Help Us Find the Right Person When She Can't?

2019-11-27

Another case of domestic violence has occurred. In the elevator, the girl resisted desperately, shouting for help, but she was no match for the strong man with a fat body and was dragged out. What happened afterwards, even without footage, we can guess. On the afternoon of November 25th, Yu Ya, a beauty blogger who became famous for imitating the Mona Lisa makeup, uploaded a video on Weibo, writing, "I was abused at home." The video quickly gained widespread attention, with over 460,000 reposts. The girl who was beaten is named Yu Ya, a beauty blogger on the internet. She has done many makeup imitations, and because the effects were really good, she even signed a contract with Papi Jiang's company. The abuser is called "Tuo Tuo," a so-called intellectual. His Weibo name is Tuo Tuo's Wind Demon Cult, and his introduction says, "No one can stop me." How did these two people meet? The story happened in July this year. Yu Ya's imitation makeup video of the Mona Lisa that she shot last year was seen by a well-known director in China who liked it and wanted to contact Yu Ya to collaborate with her. The director had limited acquaintances in Chongqing, so he asked Tuo Tuo to help find someone. Tuo Tuo contacted Yu Ya through a friend's Weibo. The three of them also created a WeChat group, and that's how Tuo Tuo and Yu Ya met. During the conversation, Tuo Tuo exaggerated his career as much as possible, claiming to be a screenwriter who has published books, art albums, and collections of works. He spoke humorously and elegantly, presenting himself as a mature and successful man. Yu Ya was deceived by this. In September, the two officially started dating. The nightmare began from there... In recent years, there have been many cases of domestic violence exposed, and even many celebrities and internet celebrities have been deeply involved. Most of the victims of domestic violence seek help voluntarily, and their experiences are only known to those around them. But why did the two people who were once in love end up hurting each other? "From now on, you can only love me, cherish me, not deceive me. You must fulfill every promise you make to me, speak to me sincerely, not bully or scold me. You must believe me and come to my aid immediately if someone else bullies me..." These vows always seem to expire easily. Finding the right person and growing old together is a difficult problem in life. Data shows that in 2019, the total number of unmarried people over the age of 15 in China was 215 million (127 million males and 88 million females), and with the addition of divorced people, the number of single people reached 240 million. The number of registered marriages has been declining since 2013, and by 2017, it had dropped to 7 per thousand. Looking at the world as a whole, the proportion of single people is 45% of the adult population in the United States, 32.4% in Japan, and 23.9% in South Korea, while in China, it is only 17.4%. Compared with developed countries, the proportion of single people in China is not high. With further economic development, the number of single people in China is likely to continue to grow and reach 400 million. Data shows that the average age of first marriage for single men and women in China is getting older. In first-tier cities, it is not uncommon for people to get married at the age of 30. The reasons behind this are that people have a fast-paced work life and high pressure, leaving little time for socializing. On the other hand, everyone hopes to find the perfect partner and grow old together, but this is not easy. In order to meet the needs of single men and women to find a partner, various marriage and dating websites and apps have emerged in recent years. Both domestically and internationally, they use big data and AI algorithms to match single men and women with their "Mr. Right." AI matchmaking apps are already very common in China, but they are monopolized by matchmaking companies. The existing marriage and dating websites and apps can be divided into several types: direct matching based on user big data, recommendation based on user information collected through chatbots, and matching based on facial recognition. How does AI perform matching? Based on neural networks, the American app "Bernie" uses machine learning, artificial intelligence, and neuroscience technologies to automatically select dating partners for users and facilitate introductory conversations. It simplifies online blind dating matching by automatically swiping through photos of people who are interested in the user and sending them date invitations. Banihal asks new users to answer more than 40 questions during registration, similar to the "36 questions that lead to love" by psychologist Arthur Aron. It uses these questions to help users gain a deeper understanding of themselves and their preferences. After users fill out comprehensive profiles, the system recommends the top 5 matches for them. Banihal uses precise algorithms and the software called Rae to ensure successful marriages. Rae can mine subconscious signals and has been upgraded with artificial intelligence and data to provide a better user experience. However, Bernie's app and data are based on Tinder, a dating app, and without reliable data sources, it is almost impossible for AI to successfully play matchmaker. In fact, these systems are similar to recommendation systems used by shopping websites like Amazon and Taobao, using machine learning-based clustering algorithms to determine similar groups of people. Similarity can be defined based on behavior (such as what they buy or browse, how they pay, purchase frequency, etc.) or characteristics (such as age, gender, location, etc.), or a combination of both. The recommendation engine predicts a user's future choices based on their past actions and also takes into account the behavior of other people who "like" similar things. A young man from the University of Waterloo used machine learning to find a partner because of the extremely unbalanced gender ratio and lack of social activities at the university. He first analyzed the factors that determine whether one can find a partner. He listed the characteristics of men to find out which hypotheses could be supported by data. He assigned a value of 1 or 0 to each characteristic based on whether it met the criteria. In other words, he was measuring the relationship between these attributes and the ability to find a partner. He also analyzed various reasons for being single, and fitness, glasses, and confidence were the three most significant variables. Fitness: People who regularly go to the gym or exercise have more than twice the chance of having a girlfriend (p-value = 0.02). Glasses: People who don't wear glasses have a 70% higher chance of having a girlfriend than those who wear glasses (p-value = 0.08). Confidence: People with confidence have a higher chance of having a girlfriend (p-value = 0.09). He then statistically analyzed the relationships between these variables, and the data showed that {having a girlfriend, appearing confident, going to the gym, not wearing glasses} were related to each other. After doing this preparation, the young man trained a logistic regression generalized linear model to predict the likelihood of finding a girlfriend based on the explanatory variables mentioned earlier. Using the glmnet and caret packages in the R language, he trained this generalized linear model with elastic net regularization. He optimized the hyperparameters using a standard grid search and used leave-one-out cross-validation in each iteration, optimizing the kappa coefficient. The final model achieved a cross-validation ROC AUC score of 0.673. This means that the model is more reliable than random guessing when it comes to predicting the likelihood of finding a girlfriend. However, it is known that four months after the young man published the article, he was still single, and it is unknown what his current situation is... Humans are social beings and cannot live in isolation from society. AI matchmaking, divorced from sociality, such as matching Anglenababy from Texas with Xiaoming from Lu'an with a 100% compatibility, is unlikely to happen. A 99.99% compatibility is likely to be defeated by that 0.01% incompatibility. Most people do not search thousands of times to find the one that is 100% suitable for them, but rather, it is due to their own reasons and the influence of their families. Moreover, these so-called data cannot detect a person's true personality and the complexity of emotions, which means that online matchmaking cannot rely entirely on AI and big data. Even with highly accurate matching, it cannot guarantee that both parties will have a 100% chemical reaction and find love. Even after falling in love, it cannot guarantee that the other person is truly the right one. Therefore, big data matching and recommendation algorithms can only be used as a means to improve efficiency. The decision-making power in emotional matters on matchmaking platforms always remains in the hands of each user. Remember, the power to make decisions in relationships always lies with each individual. Going back to the beginning of the article, if you have been subjected to domestic violence, do not tolerate it, but take legal measures to protect yourself and say no to domestic violence! We also wish all the girls who have been abused the courage to speak up, no longer remain silent, bravely defend themselves, leave the scum, and start a new life.

RECENT AI NEWS

RECENT AI TOOLS