An AI developer has found his “happily ever after” after a self-reported year-long experiment using ChatGPT and other AI tools to date more than 5,000 women on Tinder.
In a Jan. 30 thread on X, TenChat product manager Alexander Zhadan announced he’s getting married in August to artificial intelligence (AI)-endorsed soulmate Karina Imranovna after whittling down 5,239 Tinder matches using an AI system he devised.
It filtered out women according to his red flags, made small talk on his behalf, booked dates and even helped him propose to his now-fiancee.
In a 37-part thread, Zhadan revealed he started the experiment a year after his ex-girlfriend (who now claims they were just friends) ended their two-year relationship in 2021. Returning to dating, Zhadan said his Tinder relationships “went to a dead end” within weeks.
“I decided to simplify communication with girls via GPT.”
Zhadan began the project after gaining access to the GPT-3 application programming interface, or API, in 2022, explaining it was initially a simple script that would “swipe right” to women with more than two photos on their profile.
The V1 worked on the request “find the best one, keep in touch,” but at the same time, the system often forgot information, was limited to communicating on Tinder, and occasionally communicated poorly.
However, after the release of ChatGPT in late November 2022, his system got an upgrade to “V2.”
This paired ChatGPT with photo recognition software that could filter out women whose profiles had zodiac signs, explicit photos and other personalized criteria.
The V2 system divided the databases into general and personal messages, muted replies, and replies to several messages. It added photo recognition using FlutterFlow, created trust levels for sharing personal information and could write messages itself.
Chat-up GPT
The upgrade also enabled his AI bot to sound more like him when talking to his Tinder matches.
“ChatGPT found better girls and chatted longer [...] ChatGPT swiped to the right 353 profiles, 278 tags, he continued the dialogue with 160, I met with 12.”
However, the system sometimes promised his dates flowers or chocolates without his knowledge and occasionally double-booked him for dates.
Zahadan used the combination of v1 and v2 to find the most suitable date. While at first, he would ask v1 to talk about his childhood, parents, goals, and values, he later transferred the data to v2, speeding up compatibility. This system helped him find more than 4,000 matches per month on Tinder Gold.
“This time, there are 4943 matches per month on Tinder Gold, and it’s scary to count how many meetings,” said Zhadan.
Eventually, Zhadan said he limited his interactions to just four women at a time and ended up getting on “serious terms” with just one: Karina.
Related: OpenAI and Arizona State University partner for ChatGPT implementation
This prompted Zhadan to build “v3,” aimed only at interacting with Karina — though he admits it was only used to serve as a stopgap if he didn’t respond for an extended time.
Zhadan set up v3 as an observer who communicates when he cannot write for a long time and advises him on communicating better. Some of the critical commands included support, not quarreling, and offering activities. For example, when a conversation with Karina got heated, Zhadan asked v3 to maintain and improve the relationship with Karina, and the v3 system was obliged to make the conversation more positive.
The experiment turned another corner in November 2023 after ChatGPT recommended Zhadan marry Karina based on a conversation with her and even helped devise a plan for him to propose during a trip to Macau and Hong Kong.
According to a screenshot of a marriage registration booking confirmation shared by Zhadan, the pair are set to get married on Aug. 3, 2024.
The groom-to-be once made the news after admitting to using OpenAI’s ChatGPT to write his entire university thesis in 2022 and was even allowed to keep his diploma.
“The development of the project took ~120 hours and $1,432 in API. [...] If you met yourself and went on dates, then the same thing took 5+ years. [...] Thanks to ChatGPT for saving money and time,” said Zhadan.
Cointelegraph reached out to Zhadan for further comment but did not receive an immediate reply.
Magazine: Crypto+AI token picks, AGI will take ‘a long time’, Galaxy AI to 100M phones: AI Eye