Study: 38% open up to AI about relationships

A study by the Oxford Internet Institute found that 38% of regular UK users of large language models turn to AI for advice on personal relationships, and 31% use it for emotional support. The survey of 2,000 adults revealed that younger users and women are more likely to seek emotional support from AI, while men use it more for practical tasks. Researchers caution that AI lacks emotional intelligence and may pose risks for sensitive issues.

Emotional support and advice have long been difficult to come by, prompting a growing number of people to turn to AI for more help. The Oxford Internet Institute surveyed 2,000 UK adults online to learn about their interactions with large language models and found that almost one-third 31% of regular users reported using these models for emotional support, including talking through problems or making decisions. Furthermore, one quarter of regular LLM users look to these tools for meaningful conversation. This trust even extended into more sensitive topics, with 38% turning to LLMs for advice on personal relationships and two-thirds for information on health issues. The demographics of the respondents tended to skew the level of comfort and reliance on using chatbots for different tasks: - Younger users: more likely to use LLMs across all types of activity - Women: more likely than men to use LLMs for personal and emotional support - Men: turned to AI for help with practical tasks to a greater extent than women Despite the differences, the overall consensus was positive: 75% of the respondents were enthusiastic about the possible benefits of AI chatbots. It is worth acknowledging that there are risks associated with using AI for sensitive issues https://www.thedeepview.com/articles/ai-fakes-emotion-but-the-consequences-are-real that have traditionally been handled by humans, given the emotional intelligence and social skills AI cannot replicate. "Our study provides an important starting point for understanding LLM usage trends, but substantial further research is needed to explore the wider societal implications associated with LLM use for social and emotional support, including when such interactions may complement existing sources, and when they risk replacing important human relationships," said Florence Enock, Ph.D., Senior Research Fellow at the Oxford Internet Institute and lead author of the report. Existing research https://arxiv.org/abs/2504.18412 from the Human-Centered Artificial Intelligence Institute at Stanford University found that AI therapy chatbots can contribute to harmful stigma and dangerous responses and are, as a result, less efficient than human therapists. That's not even factoring in the potential privacy risks involved with giving a chatbot sensitive personal information, such as the state of your mental health. Our Deeper View Chatbots are becoming increasingly skilled at understanding natural language and mimicking the human experience with greater consistency. This makes it easy to personify or give human attributes to models that are just analyzing patterns and repeating them, which can appear to look like consciousness. Combined with the fact that AI labs are giving users free access to even their most advanced models, this creates a recipe for users to develop a dependency on AI for all sorts of tasks, including personal or sensitive ones. However, using any chatbot for assistance with mental health should be handled very carefully, both because of the privacy concerns and because of the unreliability of LLMs due to hallucinations and sycophancy.