I have been researching making a therapy chat bot for years... it's going to happen.
How long do you think it'll take ad companies to do an end run around patient protections and start serving ads based on "metadata"? Got anxiety? Try our new meds! Worried about the future? How about some life insurance or annuities? Etc. Things will get very exploitative very fast.
If they wanted to be nefarious in this way, they could just mine your chats and emails and get a great idea about your state of mind. They wouldn't even have to get around potentially legally privileged information.
My go-to related example of this: People often worry that Facebook is listening to their conversations because they'll see an ad related to a conversation soon after. You probably shouldn't worry that they might be listening to you (because they're not), but you might want to worry that they didn't need to listen to you to know what ad to serve you.
My wife and I, we experience a "Facebook is listening" event roughly every month these days. Of course I know they aren't, but these too-well-targeted ads or post recommendations aren't about generic, easy-to-guess interests - they're about completely random things that we just happened to spontaneously and briefly talk about that same day. The most likely explanation to me is that those conversation topics weren't actually spontaneous - that is, we were already primed, nudged to talk about this, and while we didn't know that, ad targeting algorithms did.
Learned about it in the documentary Century of the Self, an outstanding work.
Black Mirror was mentioned not because it pioneered this idea but because each episode shows a possible dystopian future associated with various technological advances.
These limits can and will be blown away by bots given the amount of data they can crunch. Only issue is unlike in games like Chess or Go where the end state is clearly defined, with therapy its open ended and unknown what type of end states are possible.
Just a matter of time before all paths get mapped out.
I like Google's TTS and it was incredibly annoying to integrate with my bot. I realize they don't want you using it for free for your own purposes, but damn.
It's like talking to myself, which is what I've always done and it's always the same conversations. So I automated it.
Now half+ of it comes from outside, which seems to work better even if it's a synthetic voice with messages I made.
I also used it for nagging medication reminders, as I kept forgetting what, how and when to take (at one point it was... quite a lot, on an overly complicated schedule).
It was too much emotional work to work with the training data so I haven't worked on it in a long time, but if something like that would be helpful, feel free to message me as well. :)
Or, if there's some community somewhere where people trying to solve this common problem gather, I'd love to hear about it / join!
He cured my potty mouth, THAT'S NOT MY PROBLEM