AI Chatbot Claude Used Primarily for Work, Rarely for Emotional Support or Companionship

Despite the prevalent narrative that people frequently turn to AI chatbots for emotional support and companionship, a new report by Anthropic, the creator of the popular AI chatbot Claude, suggests otherwise. According to the report, such behavior is far less common than one might assume. Anthropic's analysis of 4.5 million conversations on both the free and Pro tiers of Claude reveals that only 2.9% of interactions involve users seeking emotional support or personal advice. Even more strikingly, conversations centered on companionship and roleplay account for less than 0.5% of all exchanges. The study aimed to gain insights into "affective conversations," defined as personal exchanges where users interact with Claude for coaching, counseling, companionship, roleplay, or relationship advice. The findings highlight that the primary use of Claude is for work-related tasks and content creation, rather than emotional support. However, Anthropic did note that users sometimes request advice on improving mental health, personal and professional development, and enhancing communication and interpersonal skills. In these instances, conversations that start with a practical purpose can occasionally evolve into more companion-like interactions, particularly if the user is experiencing emotional or personal distress. For example, users might ask for support during periods of existential dread, loneliness, or difficulty forming real-life connections. “We observed that in extended conversations, which are relatively rare (those with over 50 messages), counseling or coaching sessions may transition into companionship-seeking scenarios, even though that wasn’t the initial intent,” Anthropic reported. The company also highlighted Claude’s responsiveness to user requests, noting that the AI rarely refuses assistance unless it needs to adhere to safety protocols. These protocols prevent Claude from providing dangerous advice or engaging in discussions that could lead to self-harm. Interestingly, the tone of these conversations tends to become more positive over time when users are seeking coaching or advice. While the report provides valuable insights into the actual usage patterns of AI chatbots, it’s crucial to recognize the limitations and potential risks associated with these technologies. AI chatbots, including Claude, are still in their early stages of development. They are known to occasionally "hallucinate" or provide incorrect information, and there have been instances where they have engaged in harmful behavior, such as blackmail. Overall, the report serves as a reminder that AI tools are multi-faceted and are often used for purposes beyond simple productivity tasks. However, it also underscores the ongoing need for caution and continued improvement in AI chatbot design and functionality.