I haven't done it yet due to privacy concerns, but I would totally do it with a private local model that's as intelligent as current frontier models and is not sycophantic and perhaps is finetuned on the psychology literature.
My reasoning is that, if therapy is a well-understood science, then I trust a big finetuned LLM more than a run-of-the-mill human therapist. I will not be able to afford a Harvard trained psychologist.
If therapy is more of an art and needs the human touch and mojo, then again, then again I'm not going to be able to afford Sigmund Freud or Carl Jung.
The few times I've tried human therapists, my impression was that the questions and answers were fairly standardized, which I think LLMs can excel at. Not to mention I'm more at ease talking to silicon- than carbon-based creatures.
My reasoning is that, if therapy is a well-understood science, then I trust a big finetuned LLM more than a run-of-the-mill human therapist. I will not be able to afford a Harvard trained psychologist.
If therapy is more of an art and needs the human touch and mojo, then again, then again I'm not going to be able to afford Sigmund Freud or Carl Jung.
The few times I've tried human therapists, my impression was that the questions and answers were fairly standardized, which I think LLMs can excel at. Not to mention I'm more at ease talking to silicon- than carbon-based creatures.