A study by OpenAI and the MIT Media Lab has shown that ChatGPT might be making its regular users feel lonelier. Since its launch over two years ago, ChatGPT has become very popular, with more than 400 million people using it every week. Even though it’s not meant to be a friend, some users develop strong emotional connections with the AI, which led to the research.
The researchers conducted the study in two parts. First, they reviewed millions of ChatGPT chats and audio interactions while surveying over 4,000 users about how they use the bot. Then, the MIT Media Lab recruited 1,000 people to participate in a four-week trial, during which each person used ChatGPT for at least five minutes a day.
They discovered that while many factors can lead to feelings of loneliness, users who trusted and formed a bond with ChatGPT were more likely to feel lonely and rely on AI. In short, the study found that using ChatGPT more each day was linked to increased loneliness, dependence, problematic use, and less social interaction overall.
Researchers also looked closely at how users interact with ChatGPT’s Advanced Voice Mode, a speech-to-speech interface. The chatbot could work in two ways: a neutral mode where it kept a neutral tone, and an engaging mode where it shared feelings openly.
The study found that, at first, using a voice-based chatbot helped reduce loneliness and dependence more than text-based chats. However, these benefits faded with heavy use, especially with the neutral mode.
Even though the technology is still new, the study could help spark a discussion about its effects on mental health over time. “We’re just starting the conversation about what we can measure and what the long-term impact on users might be,” said Jason Phang, an OpenAI safety researcher.
This research comes as OpenAI has released GPT-4.5, which they say is more intuitive and emotionally aware than earlier models and competitors.
You might also be interested in-Restaurant sells half chicken for ₹5,500, claims it was raised on music and milk
