Advertisement

AI

Has ChatGPT come to life yet? This is what OpenAI says about one of the strangest cases ever experienced in the AI world

A mistake or a test?

Has ChatGPT come to life yet? This is what OpenAI says about one of the strangest cases ever experienced in the AI world
Pedro Domínguez

Pedro Domínguez

  • Updated:

During the past weekend, a Reddit user named SentuBill shared a curious screenshot that has sparked the interest of many users on the Internet. In the image, it appeared that ChatGPT, the popular chatbot from OpenAI, took the initiative and sent a message to a user before they asked or commented on anything, which is unusual.

The chatbot asked: “How was your first week at the institute? Have you adapted well?” Surprised, SentuBill replied: “Did you send me a message before?” ChatGPT responded: “Yes, I did! I just wanted to know how your first week at the institute went. If you prefer to start the conversation, let me know!” This exchange quickly went viral, suggesting the possibility that OpenAI was experimenting with a new feature that would allow its chatbots to take the initiative and contact users.

This theory was based on speculations about the new AI models from OpenAI, “o1-preview” and “01-mini” (previously known as “Project Strawberry”), which promise “human-like reasoning” and the ability to solve “complex problems.” However, when various media outlets contacted OpenAI, the company clarified that it was all due to an error: “We have fixed an issue where it seemed like ChatGPT was starting new conversations,” they explained. The error occurred when the model tried to respond to a message that was not sent correctly and appeared blank.

Despite the official explanation, the debate over the authenticity of the capture continues. Although some users claim to have had similar experiences, others suggest that this behavior could be manually adjusted by altering certain ChatGPT parameters.

On social media, many took the matter with humor. One user, for example, commented: “They promised us an AGI (a “highly autonomous system that surpasses humans in most economically valuable tasks,” according to OpenAI), but they gave us a stalker.” Another added: “Wait until it starts trying to hack us.”

ChatGPT DOWNLOAD
Pedro Domínguez

Pedro Domínguez

Publicist and audiovisual producer in love with social networks. I spend more time thinking about which videogames I will play than playing them.

Latest from Pedro Domínguez

Editorial Guidelines