I found my therapist through ChatGPT

I was surprised when I heard my client say this yesterday.

I was meeting her for the first time. She's a teenager, and on our introductory call with her parents she said she found me online. Being the Gen-Z-er that I am, I assumed she meant Google. Instead, she pulled out her phone saying, "Hey, can I show you my chats with ChatGPT? I've been using it as a place to vent and it told me I should probably talk to a therapist - and then it suggested you."

After the initial surprise, I swiftly moved on to reading her chat - which more or less read like a diary entry. The session proceeded with, I am somewhat glad to say, no further input from ChatGPT.

However, I've been mulling over the incident for a day now. Several things about it surprised, and in a strange way, reassured me. In the first place, I was surprised that my client, a young teen, was already so familiar with ChatGPT. How was she introduced to it? How often had she used it? For how long had she been confiding in it? All questions I have not asked her yet - and which I may never ask if the work does not call for them.

The other thing that really surprised me was ChatGPT's recommendation that she speak to a real, live therapist. It is heartening to know that AI is recommending that people talk to other people. That the developers have created a way for the system to know when it is out of its depth.

I haven't tried out ChatGPT as a confidant; I doubt I'll ever feel comfortable putting my deepest insecurities and most troubling worries out onto the internet like that. Unfiltered, almost like a diary - except it's being read by a system that then can use that information in unpredictable ways for its own ends.

However, having seen my client use it, and knowing that plenty of other people do the same, I'm glad to know that there are guardrails built into ChatGPT that recommend professional help. Well, at least they make sure that I'm not going to run out of work anytime soon! Jokes apart, though - it really is important to have safeguards, and I'm glad to see that in this instance at least, they seem to have worked.