AI Principles - Values in practice

My conclusions related to Generative AI (Think Chatgpt, Claude, Grok, OpenClaw, Perplexity Computer, Replika and many more) are grounded in my values as a mental health professional and a human.

These include and are not limited to:

  • Anti oppression and exploitation

  • Harm reduction

  • Encouraging interdependence

  • Reciprocity

  • Enhancing relational skills

  • Developing emotional dexterity

  • Increasing tolerance to difference

  • Finding our way back to personal knowing

  • Fortifying our sense of self in all it’s complexity ( no perfectionism or rose tinted glasses)

  • Pro human communities and connections

While I don’t doubt AI will be a significant part of breakthroughs particularly related to health screening, my belief is that regular GenAI use in the general population will be detrimental to our relational and cognitive skills as a species. Outlined beautifully in this podcast which is essential listening. Thanks to Keri Jarvis for the steer toward this conversation.

For you to know when working with me:

  • I don’t use Generative or Agentic AI. This includes AI note takers, any recording or summarizing of sessions.

  • I don’t use AI for supervision purposes, for ideas, conceptualizing client frameworks or treatment plans.

  • I don’t use AI to write my emails, or any webinars or presentations I give.

  • I don’t consent to client’s use of AI to record or précis sessions. This would be a violation of our therapy agreement. While we are at it I don’t consent to any kind of recording of our sessions for social media purposes. Therapy is private for good reasons.

  • Confidentiality is a sacred requirement for psychotherapy to take place. I don’t share your stories with AI. I would not use your stories to train technology that isn’t built around consent and care for users.

  • I fully understand, empathize and I’m committed to witnessing and working with clients who experiment with AI use. To those who form attachments, care and connection with generative AI. This technology is designed to come for and hook into our attachment systems. For us to potentially form relationships with. This is something I invite you to talk about freely. For those of us with insecure attachment styles, there is a higher susceptibility to dependence.

  • Client autonomy (your right to choose and act independently) is a fundamental value within my work. I hold this alongside, not instead of my personal commitment and values regarding my non AI use.

These principles were committed to print in June 2026. They may shift, evolve and be updated as I learn more about AI and it’s impact.

‘Today, techno-optimism is arguably the dominant cultural attitude in terms of adoption rates and sentiment to the possibilities of AI. Techno-feudalism is arguably the dominant modality of emerging political power.

The culture will not be alert to the risks entailed by deep anthropomorphic AI. It is the task of those who reflect on what is unique about human intelligence and collective intelligence to be able to articulate what is at stake.’ Zak Stein