Neural network helped him find a mate by communicating with girls instead of a Muscovite for a year

No time to read?
Get a summary

Alexander Zhadan, a young man from Moscow, created the ChatGPT chatbot to search for dates and communicate with girls. So, for a year the neural network met 5,240 girls, and a marriage proposal to one of them was made by a man. It’s about Zhadan said on X (Twitter).

Zhadan started the experiment in 2022. It used Tinder and the GPT-3 language model. The neural network was given a task: to meet a girl and ask her out, but not right away. According to the enthusiast, the first version of the service turned out to be not very effective. He engaged in less-than-convincing dialogue, forgot the context, and immediately offered to meet.

Later, Zhadan redesigned the service around GPT-4. The neural network began crafting more organic responses and also learned to show the guy potentially wrong messages before sending them. As a result, the new version of the complex gained 4,943 “matches” (mutual likes) in a month and made nearly a hundred appointments. The number of daily appointments reached five.

“On dates we ate and drank at a bar, then watched a movie or walked the streets, visiting exhibitions and tea houses. It took 1-3 meetings to find out if it was the one. I understand that people often meet differently, but for me this process was very different,” Zhadan said.

The “same” girl turned out to be a certain Karina. Zhadan proposed to her on January 30, 2024.

Early demand for IT specialists in Russia augmented doubled.

No time to read?
Get a summary
Previous Article

Winners of Benidorm Fest television voting Mantra: “We have a bittersweet taste”

Next Article

Leszczyna threatens hospitals with breaking contracts with the National Health Fund! “The manager of the facility must ensure termination of pregnancy in accordance with Polish law”