The Bothub platform conducted a comparative test of five AI models—ChatGPT, Claude, Gemini, DeepSeek, and Grok—to evaluate their effectiveness as tools for psychological self-help. The test was based on a specific user request regarding insomnia caused by ruminating on a stressful conversation with a supervisor.
ChatGPT, Gemini, and DeepSeek were found to be more focused on providing practical techniques and structured advice. In contrast, Claude and Grok demonstrated a stronger emphasis on dialogue, with Claude offering deeper analysis and Grok communicating in a more informal, friendly manner.
The platform provides access to these models through a single interface, eliminating the need for multiple subscriptions. Users can pay from a single balance in rubles without the need for a VPN or linking a bank card. New users are offered 300,000 bonus tokens to test the service.
The authors provided a specialized prompt designed to guide AI models to act as experienced psychologists with a 15-year background. This prompt instructs the AI to avoid clichés, focus on active listening, and refrain from diagnosing users, while emphasizing the need to refer individuals with severe conditions to human professionals.
