I agree. The more I think about this the more cautious I become. I do think there will be good evolutions in the space, but I also think it’s best taken on by someone who has access to the safeguarding capacity to really ensure it does no harm; perhaps a forward thinking university or therapy organisation.

For now, I’m leaving this firmly in the, ‘would be awesome, but not for me’ list.

My gut feel is AI based therapy risks scoring very lowly on your ‘Do No Harm’ metric if, for example, someone sought therapy from your tool and then did themselves or others harm off the back of the advice. A human is far better at reading between the lines, and understanding cause and effect of what they say/do (and how) than an AI is, for now.

I agree Roberto – that’s a good idea. I’ve added it to my shortlist.

If I was to approach that it’d be from a chatbot/coach stance rather than a therapist. I think there is much good to be done with AI therapy, but I recognise that it’s life-or-death stuff sometimes, and I think it’d need serious serious investment to do it well.

One of my key requirements for this project is “first, do no harm” – and I’d worry that a bug or an AI hallucination might cause harm if I sold something as a therapist.

Going to think on this more though, thanks for the comment/idea.

Are you thinking of making something along those lines too?