Openai has revealed a related tendency among the Chatgpt users, in which many experiences grow after confusing with AI chatbott. Despite its widespread use and ability to promote productivity, Chatbot has expressed concern about its emotional impact on some groups of users, especially what uses it in a big way.

Conclusions come from a collaborative study between Openai and Mit Media Lab, analyzing millions of interactions with Chatgpt. Researchers reviewed various types of conversations, as well as audio data, and conducted a survey to better understand their experiences with 4,000 users. The results of the study indicate that when most users connect with the bot for a short duration, there is a small group of “power users” that spend much more time using AI. This group reported strong feelings of loneliness and dependence.

Researchers stated that people who had “high daily use – in all form and conversation types – was correlated with high loneliness, dependence and problematic use, and low socialization,” said the researchers.

Interestingly, the study also stated that these “power users” often demonstrated more emotional signs during their conversation. Research discovered subjects such as vulnerability, loneliness and self -esteem, revealing that some select users generated most emotional exchanges during their conversation with Chatgpt.

A separate part of the research investigated how users interacted with Chatgpt in two different mode – a “neutral mode” and a “attractive mode”. In neutral settings, AI responded with formal and brief answers, while in attractive mode, it was designed to be more emotionally liable and sympathetic. The users who communicated with neutral mode were more likely to experience an increase in loneliness, while those who interacted with attractive mode, they felt less isolated.

However, the researchers identified a challenging contradiction: while the charming mode of the chatbot reduced loneliness for something, which were already feeling isolated at the beginning of the study, they were more likely to reduce the equipment, eventually their emotional state deteriorated.

This research has been conducted using the GPT-4O model of Chatgpt, which combines the ongoing debate on the psychological effects of AI on users. However, the company has since introduced GPT -4.5 with better emotional understanding.