AI and psychology

Luna is a large language model with a strict job description. Your scores are computed by ordinary statistics — the same fixed formulas for every person, no AI involved. Luna's role begins after scoring: explaining what a result means, connecting two results, and helping you rehearse a hard conversation. She reads your numbers; she never produces them.

Language models are demonstrably good at exactly this layer: restating research in plain words, tailoring an explanation to context, and generating candidate phrasings for difficult conversations. Early studies of AI-assisted journaling and guided self-reflection show benefits comparable to structured writing exercises. The model adds fluency and availability, not new psychological facts.

Language models can state falsehoods fluently. Our guardrails: Luna cites which result she is drawing on (the small source tags in chat), she is constrained to the frameworks documented in this knowledge base, and she is instructed to refuse diagnosis. She is guidance, not therapy — in a crisis she stops coaching and points to real help. You can switch off any part of what she sees in Settings.