Contents:
Free Course: "Quick Start in Python"
Learn MoreAlexander Zhadan gained popularity across the country thanks to a Twitter thread in which he described how he wrote a thesis in 24 hours using ChatGPT. A year later, he shared a new, captivating story about how a neural network helped him find love. Over the course of a year and a half, his algorithm interacted with 5,239 women on the dating service. One of his successful encounters was Karina Vyalshakaeva, who later became his fiancée. This story highlights how modern technology can influence personal relationships and help people find each other in the world of online dating.
In an interview with Skillbox Media, Alexander shared his motivations for conducting the experiment, talked about the team that helped him develop the algorithm, and expressed his point of view on accusations of unethical behavior.
Tired of endless swipes and correspondence
The idea for a dating chatbot arose in response to users' growing interest in online interactions. With the development of technology and the increase in people searching for partners online, the need for convenient and effective tools has arisen. Chatbots can significantly simplify the dating process, offering users the opportunity to communicate and find suitable partners at any time. Importantly, this approach makes the process less formal and more interactive, which attracts users looking to make new connections.
The idea for the project arose in the summer of 2022, when I began exploring the GPT API. At that time, ChatGPT didn't exist yet. I was tired of endless swiping and messaging on dating apps, which often led to the same formulaic conversations and yielded no results. That's why I decided to use the power of artificial intelligence to try to find the right person for communication and further relationships.
How long did it take to develop the algorithm from the initial idea to its more or less working version?
The project was created over about a year and a half. Initially, I developed it for myself, but over time, I wanted to share the results with others.
I have experience in the field of neural networks and am actively involved in their development. My background includes training and practical application of various machine learning algorithms, as well as working with modern frameworks for creating and optimizing neural networks. Currently, I am focused on the application of neural networks in various fields, such as data analysis, image processing, and natural language. This allows me to stay at the cutting edge of technology and develop innovative solutions to current problems.
I completed my studies at the Russian State University for the Humanities (RSUH) in 2023, receiving a degree in Management. Defending my diploma using ChatGPT brought me a certain fame, perhaps even on a national level.
Read also:
The author of a thesis created using artificial intelligence shared his motivations for revealing this story. He believes it is important to highlight the process of creating academic work using technology to demonstrate how modern tools can change the approach to teaching and research. This also highlights the need to discuss the ethical aspects of using AI in the educational environment. The author hopes that his experience will help other students and teachers better understand how to effectively integrate technology into the educational process.
Although I studied at a humanities university, I was actively involved in game development during my studies. I focused primarily on web services, studying code and analyzing the operation of various functions. As part of public projects, I collaborated with a team developing an AI psychologist. This experience allowed me to deeper understand aspects of game development and the integration of artificial intelligence into modern applications.
Currently, I hold the position of Neural Networks Project Manager at TenChat. As part of my work, I have also implemented several projects related to artificial intelligence. One such project was brand media optimization for M.Video. We leveraged ChatGPT and, to some extent, Midjourney to create high-quality materials and improve content.
Various machine learning methods and technologies were used to train the artificial intelligence used in Tinder. The primary tool was data processing algorithms that analyze user preferences and behavior. These algorithms collect information about which profiles attract users, which messages receive the most responses, and general trends in app usage. Neural networks were also used to recognize patterns and improve recommendations. As a result, this integrated approach allows for more accurate and personalized matches, increasing the effectiveness of user interactions in the app.
I used sanitized dialogue from Tinder to train the language model. When ChatGPT, for example, had to answer a question about my school, it consulted a table of answers rather than its own database. As a result, he would formulate answers in his own style, for example, "I'm a fifth-year student at the Russian State University for the Humanities."
At first, the results were unsatisfactory: although GPT could generate answers, they were unnatural and uninspiring. I later optimized the system to take into account criteria important to me and to offer more personalized answers.
The technological foundation of this model includes the use of modern libraries and tools that ensure high performance and efficiency. During development, libraries such as TensorFlow and PyTorch, which provide powerful capabilities for working with neural networks, were used. Data processing libraries such as Pandas and NumPy, which allow for the efficient manipulation and analysis of large volumes of data, were also used. Matplotlib and Seaborn were also used to visualize the results. Together, these tools create a robust and flexible environment for model development and deployment, delivering the optimal balance of speed and accuracy.
The platform hosts numerous repositories, including Torchvision, Thunders, Alchemy, Selenium, Pandas, LangChain, FlutterFlow, and the Chrome Driver used for Tinder. These tools provide a wide range of development and automation capabilities, making them indispensable for developers and researchers.
Initially, the project relied on GPT-3, which was significantly inferior to GPT-4 in solving creative tasks. The model works as follows: you provide a set of messages that describe your personality, and the model generates its answers based on this data. To more accurately answer specialized questions, I integrated a Telegram bot that sends notifications, allowing me to manually adjust the answers. This improves the quality of interaction and increases the accuracy of the information provided.
"The bot responded super-nicely; I don't communicate like that in messages."
My work effectiveness can be assessed by several criteria, such as the quality of tasks completed, meeting deadlines, client satisfaction, and results achieved. I regularly analyze my achievements and strive for continuous improvement. Feedback from clients and colleagues also helps me identify strengths and weaknesses so I can adjust my work and improve its effectiveness.
This project has truly proven successful. The program analyzed over 14,000 questionnaires and selected over 5,000 of them. Over the course of the experiment, I conducted approximately 100 meetings. This volume of work confirms the project's high effectiveness and significance.
This looks like a promising startup. Perhaps it's worth considering creating a separate application to implement this idea.
Creating a mass product based on this program is a complex task. This program was developed hastily and exclusively for my needs. Adapting it for other users would require going back to the beginning and starting the process from scratch. Despite the existing interest in the project, its implementation is proving unprofitable from a cost perspective. In total, approximately 200,000 rubles were spent on the project.
You mentioned that you used the questionnaire database to help companies find employees, which allowed you to cover your expenses. Can you tell me how this system was organized?
I used databases and ChatGPT to search for suitable candidates for various vacancies. After finding relevant people, I sent recommendations to the employer. If the candidate met the requirements, I received a percentage of the transaction. The maximum income was about 80,000 rubles when I helped find a senior developer.
I've heard IT specialists say that implementing such a project alone in such a short timeframe is practically impossible. Did you receive support during the development process?
I'm not trying to appear as an expert; this project was the result of teamwork. I had no previous experience with similar projects, and I received support from specialists from MMSL, as well as acquaintances from Tinkoff and MTS AI, and developers from smaller studios. There were many talented people on our team, and their names could fill a whole page of credits.
A total of a certain number of people were involved in the project. This team included specialists in various fields, which allowed us to effectively complete the assigned tasks and achieve high results.
Unfortunately, I cannot give a precise answer to this question.
Have you ever experienced artificial intelligence creating an image in communication with girls that is radically different from your true self?
I can say that it did not make me look stupid. However, the language model's communication style was different from mine. The bot sometimes used a very sweet style, while I do not communicate in such a tone. In terms of the quality of responses, it should be noted that the second version of the project began to produce more accurate and correct answers.
After each meeting, I wrote down my impressions, noting what I liked and what I did not, and whether it was worth continuing to interact with this person. Gradually, I realized that this process was becoming endless. As a result, I began to significantly narrow the field of candidates until I settled on one. That's how I met Karina, to whom I recently proposed.
She was surprised to learn that she had communicated with a neural network before we met. This fact aroused her interest and many questions about how exactly a neural network interacts with people and the extent to which it can understand their emotions and thoughts. She noted that this discovery emphasizes the uniqueness of our communication and adds a new level to our interaction.
She was completely shocked. She asked many questions about how the neural network functions, where and when I used it. I explained all aspects to her in detail. It is important to note that I always answered the key emotional moments in our communication myself. When we met face to face, all doubts disappeared, and the situation became clear.
Live communication is still a priority
Under your thread, one user noted that you demonstrate a consumerist attitude towards women. How do you respond to such criticism?
People may be surprised by an unconventional approach to dating. Society has established patterns of communication and dating. While I also adhere to a certain structure, in my case, I sometimes use a neural network. This allows me to significantly reduce the time spent on dating, as I immediately eliminate those with no prospects for a future relationship. This method makes the process more efficient and focused, which is especially valuable in today's world, where time is of the essence.
Meeting people through neural networks can indeed be perceived as devaluing communication. The question is how deep and sincere such interactions can be. If we use technology to find like-minded people or expand our social circle, this can be beneficial. However, it's important to remember that the true value of communication lies in human interaction, empathy, and the ability to understand each other. Neural networks can serve as an auxiliary tool, but they cannot replace face-to-face communication, which forms genuine connections and emotions. Therefore, it's worth considering how technology can help establish connections without losing the value of communication itself.
Neural networks are a modern tool, similar to the mobile phone we use for messaging. It's important to note that the primary significance of these technologies lies in how they impact people's lives and how they evoke feelings in users. Neural networks open up new possibilities in various fields, including communication, data analysis, and process automation, making them indispensable in the modern world.
Regarding Karina, our relationship is developing in a way that I feel genuine joy, and she shares my feelings. The fact that we met through a neural network does not diminish the quality of our relationship. On the contrary, it highlights how modern technology can help create deep connections between people.
Some people think your An act is unethical. This may be due to different moral standards and values they adhere to. Each person has a unique perspective on a situation, and sometimes these perspectives can differ significantly. Furthermore, public opinion and cultural traditions can influence how people perceive the ethics of actions. It's important to understand that ethical assessments are subjective, and different people may perceive the same action differently based on their experiences and beliefs.
I share my experience and do not insist that anyone follow my path. Everyone decides for themselves whether to accept my example or not.
Public perception of technology is always accompanied by fears and doubts at every stage of progress. However, over time, society adapts to change. Live communication between people remains an essential element of human life and consistently retains its significance, regardless of technological innovations.
Learn more about coding and programming in our Telegram channel. Subscribe to stay up-to-date with interesting content and helpful tips!
Our blog is packed with interesting and helpful content. We offer articles that will help you understand a variety of topics and expand your knowledge. Here you'll find information on the most pressing issues, as well as advice and recommendations from experts. Don't miss out on our content, which will help you stay on top of the latest trends and news. Subscribe to updates and always stay one step ahead. Read also:
- Neural networks are out of their minds: AI schizophrenics and mentally ill chatbots
- ChatGPT - what is it and how a neural network works: capabilities and examples of use
- What is a neural network and how does it work?
