Many people are turning to AI chatbots for personal needs, whether companionship or emotional support. However, in recent months, there have been heightened concerns about the potential harms of using this technology over human interactions.
According to new research from OpenAI in partnership with the Massachusetts Institute of Technology, higher use of chatbots like ChatGPT may correspond with increased loneliness and less time spent socializing with other people. The analysis evaluated how AI chat platforms shape users’ emotional well-being and behaviors through two paired studies conducted by researchers at the organizations. The studies have not yet been peer-reviewed.
Also: Microsoft’s new AI agents aim to help security pros combat the latest threats
In study one, the OpenAI team conducted a “large-scale, automated analysis of nearly 40 million ChatGPT interactions without human involvement to ensure user privacy.” The study combined this research data with targeted user surveys to gain insight into real-world applications. The users’ self-reported opinions about ChatGPT analyzed in conjunction with user conversations helped to evaluate “affective use patterns”.
<!–>
The second study, conducted by the MIT Media Lab team, performed a Randomized Controlled Trial with about 1,000 participants who used ChatGPT over a month. This Institutional Review Board-approved controlled study was designed to pinpoint insights into how specific platform features and types of interaction might impact users’ self-reported psycho-social and emotional well-being, underscoring “loneliness, social interactions with real people, emotional dependence on the AI chatbot, and problematic use of AI.”
In both studies, participants had a wide range of experiences using ChatGPT in the past. They were randomly assigned either a text-only version of the platform or one of two voice-based options for at least five minutes daily. Some participants were instructed to have non-specific, open-ended chats, while others were informed to have personal or non-personal conversations with the chatbot.
Also: Anthropic adds web search to its Claude chatbot – here’s how to try it
The overall findings uncovered that heavy users of ChatGPT were more trusting of the chatbot and were more likely to feel lonelier and emotionally dependent on the service.
However, the user outcomes were influenced by personal factors, such as individuals’ emotional needs, perceptions of AI, and duration of usage. For example, people who tended to get emotionally attached to human relationships and viewed AI as a friend were likelier to experience negative outcomes from chatbot use.
“More personal conversation types which included more emotional expression from both the user and model compared to non-personal conversations – were associated with higher levels of loneliness but lower emotional dependence and problematic use at moderate usage levels,” the researchers observed. However, “non-personal conversations tended to increase emotional dependence, especially with heavy usage.”
According to the second study, users engaging with ChatGPT via the text-only option exhibited “more affective cues” in conversations compared to voice-based users. A more “engaging voice” did not lead to increased negative outcomes for users during the study compared to neutral voice or text-based interactions. The researchers also found that very few people use ChatGPT for emotional conversations.
Also: The best AI chatbots of 2025: ChatGPT, Copilot, and notable alternatives
The researchers said the findings from both studies are a “critical first step” in understanding the impact of AI models on the human experience and will encourage more research in industry and academia.
“We advise against generalizing the results because doing so may obscure the nuanced findings highlighting the non-uniform, complex interactions between people and AI systems,” the researchers said.
Want more stories about AI? Sign up for Innovation, our weekly newsletter.
