From personal experience, some of it was just lack of discretion - no one else to talk to about this or that, and AI is/was a readily available receptacle for whatever was bothering me or I wanted to try talking about (I just remembered I had some…inappropriate chats with an AI on an account since scrubbed and deleted. This service claims to be more careful about deleting things but who really knows. Nothing I can do now, I just need to live with it, be more careful in the future, and maybe try a data deletion request).
For the more “therapeutic” uses, it’s not about personal info like names and birthdays, but details about your life and thoughts. The common example of using AI as a therapist. I’ve chatted about medical stuff, for example. Sure you CAN say "my friend has [insert generic problem here] and get the same generic advice a google search can give you. It just isn’t the same as hashing something out in a personal conversation.
I know I’m far from the only person in this position, so I guess I can serve as an example.
Hell my last chat with GPT before giving up on it was to blatantly paste in a personal vent post about something and ask it to analyze it.
Granted I’m…a lot more conscious and careful now. Idk if my laptop will run something local but I was going to give Alpaca a shot since llamafile won’t work. I’m thinking of giving that “non-account via TOR” setup a try for public services, just for the heck of it.