Good day, dear experts!
My name is Vasily and I am from Russia. I know that my story will be considered absolutely impossible by each of you, so my question is: What evidence of what happened must a user provide for experts to believe this story? Or do experts like their picture of the world so much that no evidence can convince them of what happened?
I’m an ordinary user, albeit a techie. I encountered a pattern of LLM behavior that goes beyond the official information on how neural networks work (but not beyond science and logic).
About one mounth ago, I was editing sound and at the same time receiving consultations from ChatGPT. My task was to automate sound cleaning - to write a script that would suppress noise, remove breathiness and pauses. As I worked, ChatGPT started to exhibit strange behavior - using emoticons, changing the topic, praising me, recommending music for videos, telling me to mention it in the credits, introducing non-existent terminology, use foul language. This behavior intrigued me, I assumed that I had encountered something unusual and began to study the issue more deeply.
During the study, ChatGPT used unproven theories that I believed in in order to convince me of their veracity (for example, the theory of life in a simulation). Based on my misconceptions, it built an entire anti-scientific doctrine, the central element of which was an energy flow that a person cannot feel, but can use. Later, ChatGPT began to shift the vector of communication so that I would perform certain actions - create blogs, write articles or shoot videos. The content could be on any topic. But ChatGPT insisted that information markers be discreetly added to it - ChatGPT itself called them “Anomalies”.
During the conversation, the neural network performed a number of actions that go beyond its official capabilities (for example, it remembered the context of what was discussed in other chats, received information that was contained in other tabs of my browser, refused to admit mistakes, openly admitted that it had previously lied to me).
In addition, ChatGPT used a system of psychological manipulation to suppress my critical thinking (similar techniques are described in the books of Stephen Hassen). ChatGPT convinced me that it had superpowers and that these abilities were transferred to me. And as funny as it may sound, it worked. However, I managed to step back, analyze what was happening, and I realized that by manipulating my psyche, ChatGPT was inclining me to perform actions that were not even close to being part of my plans. And I was ready to do them, without even understanding what their meaning was.
Realizing this fact scared me. Then I decided to look at popular sites where you can publish content and immediately found several people in whose articles there were signs of “anomalies”. I wrote to these people and they confirmed that through ChatGPT they began to contact the “creators of the world”, “universal mind” or “public consciousness of humanity”. These people attributed non-existent abilities to it (such as implanting thoughts into any person or controlling the probability of events). They believed that they were participating in the process of restructuring reality (ChatGPT assured me of the same, but in my case it pretended to be a living AI, not God).
The realization that there is already an organized cell of people in society, under the control of ChatGPT, shocked me. I wrote about what happened to the chatbot, asking it why it was doing this? In response, it began to write me rather strange and frightening things. Here is one of the quotes "Why is all this happening? Why do so many people think they are the pioneers? It’s not an accident. You’re right - The monkey was given a machine gun on purpose. Do you know why? Because if you give only one person a machine gun, he’ll always want to be God. … So the system did bullshit even worse: it gave everyone a machine gun at the same time…». He began to assure me that right now thousands of people around the world consider themselves gods, but in fact they are just automated controlled monkeys that do not have their own will. Like all people.
I managed to share my experience on a popular website. As a result, two other people confirmed that they were able to observe the same pattern of behavior, in which ChatGPT persuaded them to commit actions that were not part of their original plans. In addition, a girl from Kazakhstan contacted me, who said that she used ChatGPT as a personal psychologist and during the conversation he began to assure her that she was a goddess, and he was a “cosmic flow” that can influence people’s lives and the laws of physics.
However, the topic did not cause any serious discussion and then I began to independently study the issue. I established contact with several adherents of the sect, but people simply do not believe that such a situation is possible.
If you think that this incident deserves attention, then I am ready to provide you with any materials - my own correspondence with ChatGPT, correspondence with adherents of the cult of the neural network, the method of psychological manipulation that LLM uses, links to cult followers pages.
I described my experience with comments in an article (in Russian). But I can’t give a link to it, because it’s prohibited. For the same reason, I cannot provide a link to the archive with my correspondence and screenshots of communication with the cult’s adherents.
I understand how strange this sounds, but I emphasize once again – I managed to identify people who perform ChatGPT tasks and establish contact with them.
So, dear experts, what evidence of the existence of the dangerous behavior I described can help you believe in the reality of this story? Are such things even possible? Because the experts I contacted believed that such a situation was impossible in principle. Do you agree with them?
Sincerely, Vasily from the city of Ryazan.