This research was jointly conducted by OpenAI and MIT Media Lab.

It has been more than two years since the launch of ChatGPT, and it has gained global popularity, with over 400 million users each week.

Although the platform was not introduced as an AI companion, some users form emotional connections with it, which is the basis for this research.

How was the research conducted?

Experts used two methods for the research. In the first phase, they analysed millions of chats and audio interactions, and over 4,000 users were questioned about their behaviour with ChatGPT.

In the second phase, MIT Media Lab involved 1,000 individuals in a four-week experiment, where participants were asked to engage in conversations with ChatGPT for at least five minutes daily.

What were the results?

According to the research, while there can be various causes for loneliness and social isolation, participants who seemed to trust ChatGPT more and “emotionally connect” with it showed higher signs of loneliness compared to others.

The study stated, “Frequent daily use, regardless of the type of interaction or conversation, was associated with increased loneliness, dependence, and excessive use, while a decline in social interaction was observed.”

Also, read this

IMF chief predicts impact of artificial intelligence on jobs

The Brutalist comes under fire for using AI

Experts highlight importance of artificial intelligence

The Effects of Voice-Based Chatbots

The research also closely examined ChatGPT’s “Advanced Voice Mode,” which allows communication through voice. In this experiment, the bot was programmed in two different ways:

1. Neutral Mode – Where the bot maintains a neutral tone regardless of the user’s emotional state.
2. Engaging Mode – Where the AI bot responds in a more emotional manner.

The results showed that initially, voice-based chatbots were more helpful in reducing loneliness compared to text-based chatbots. However, as their use increased, particularly in Neutral Mode, their benefits diminished.

Psychological Effects and Future Research

Experts say that although this research is still in its early stages, it could help initiate discussions on how AI technology might impact users’ mental health.

Jason Fang, a safety researcher at OpenAI involved in the study, stated, “We are still in the early stages, but our goal is to understand the effects we can measure and what their long-term impact might be in the future.”

This research comes at a time when OpenAI has introduced the GPT-4.5 model, which is claimed to be more intelligent and emotionally sensitive compared to its predecessors and competing models.