A project called Here For You For Them has opened a waitlist for what it describes as an emotional wellbeing dataset designed specifically for AI training. The goal: give AI systems better data for handling mental health and emotional support conversations, an area where most large language models are notoriously clumsy.
Details are sparse. The waitlist page does not specify the dataset's size, sourcing methodology, or licensing terms. What is clear is the intent. Current AI models are trained on internet text that skews toward information delivery, not empathetic conversation. When someone tells ChatGPT or Claude they are having a hard day, the response is usually competent but clinical. A purpose-built dataset could close that gap.
The challenge is enormous. Mental health conversations require cultural sensitivity, clinical accuracy, and an understanding of when AI should step back and recommend a human professional. Getting any of those wrong is not just a bad user experience. It is a safety risk. Previous attempts to deploy AI in mental health contexts, like the Koko experiment in 2023, drew sharp criticism for testing on vulnerable users without clear consent.
Anyone working on AI-powered support tools or customer experience platforms should watch this space, but with appropriate skepticism until the team publishes details on data governance, consent, and clinical review.