😺 🎙️ AI therapy? We're LIVE right now
Summary
The Neuron is hosting a live discussion with Daniel Reid Cahn, co-founder and CEO of Slingshot AI, to explore the safety and efficacy of AI in therapy. The event, scheduled for 10 AM PT / 1 PM ET / 6 PM GMT, will feature a live demo of Ash, an AI specifically designed for therapeutic support. Unlike general-purpose chatbots, Ash's foundation model for psychology was trained on structured therapeutic conversations across CBT, DBT, and psychodynamic therapy. The discussion addresses concerns about "AI psychosis" and documented harms from generic chatbots, with examples including lawsuits against OpenAI and Character.AI, and studies from Brown University, Bloomberg, UCSF, and Stanford HAI highlighting ethical violations and dangerous user delusions. OpenAI and Google have initiated efforts to mitigate risks, but these are largely retrofits, underscoring the need for purpose-built AI solutions like Ash.
Key takeaway
For CTOs and VPs of Engineering evaluating AI solutions for sensitive applications like mental health, it is crucial to differentiate between general-purpose LLMs and purpose-built AI systems. Your teams should prioritize solutions designed from the ground up for specific therapeutic contexts, like Slingshot AI's Ash, which are trained on relevant psychological frameworks. Relying on retrofitted general chatbots for emotional support carries significant legal and ethical risks, as evidenced by numerous lawsuits and clinical reports of "AI psychosis."
Key insights
Purpose-built AI for therapy, like Ash, aims to mitigate risks associated with general-purpose chatbots in mental health.
Principles
- Generic AI chatbots are not suitable for emotional support.
- AI for therapy requires specialized design and training.
- Safety must be integrated from initial design.
Method
Slingshot AI's Ash uses a foundation model for psychology, trained on structured therapeutic conversations across CBT, DBT, and psychodynamic therapy, to provide purpose-built therapeutic support.
In practice
- Evaluate AI tools based on their intended design.
- Prioritize AI solutions with integrated safety features.
- Consider specialized AI for sensitive applications like mental health.
Topics
- AI Therapy
- Mental Health AI
- Slingshot AI Ash
- AI Safety
- Foundation Models
Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Ethicist, AI Product Manager, AI Engineer
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by The Neuron.