Please for the Love of God Stop Building AI Therapy Chatbots | Hacker News Reader

Please for the Love of God Stop Building AI Therapy Chatbots

blogtherapy.substack.com

15 pointsrpastuszak9 comments

Counter Argument:

Talking to therapists is difficult, time-consuming, painful, annoying, and expensive. The number of hoops you have to jump through just to get to the starting line is crazy, even if you are willing and able to pay the few hundred dollars an hour out of pocket yourself.

Therapy isn't like confession or church or a religious experience. Its proposed benefits will never be life changing, it will not build you up, it will not give you anything you didn't already have, and you have to pay for it.

If I can work out a few kinks in my psyche on my own I'll do that. If I bounce some words off of a convenient lie bot, then fine.

The only way I can get more f'd in the head would be to start killing, and at this point I just don't have the drive to take on murder as a new hobby.

The ancients have us covered, much of the surviving texts deal with strife, trouble , the mixed joys of life, and the inevitable contradictions that are still our lot in life and the ancients would very much recognise the horrible possibilities in taking your thoughts and reason to a souless mechanical thing that has been conjured up by powerfull forces acting in there own interests but the pit, has it's own allure and one more lesser demon is only that, one more. no shortage of the corporeal type either.

Your interesting comment immediately passed my test for "this was written by a human, not an AI" but got me to thinking about PARRY ...

  PARRY was written in 1972 by psychiatrist Kenneth Colby, 
  then at Stanford University.
  While ELIZA was a simulation of a Rogerian therapist, 
  PARRY attempted to simulate a person with paranoid schizophrenia.

-- wikipedia:PARRY [1]

One of the fascinating things that was done was to have Parry & Eliza chat together.

So it immediately occurred to me that your comment would make an excellent Lenna[2] text to feed to AI chatbot therapists right after their initial pleasantries when they ask ~ "Now, what seems to be your problem?"

The various chatbot results would then be like the headliner in a suite of fitness tests for them and so, like the Turing Test, could be named in your honor, maybe called the Bizarro Test [Suite Protocol or BTSP].

Perhaps you would help humanity in its likely brief and futile resistance by granting permission ala Creative Commons for such use?

[1] blockquote is from wikipedia:PARRY ...

https://en.wikipedia.org/wiki/PARRY

[2] "Lenna (or Lena) is a standard test image used in the field of digital image processing, starting in 1973." -- wikipedia:Lenna ...

https://en.wikipedia.org/wiki/Lenna

I don't think my reply qualifies as a standard.

It's born out of my very real frustration at finally getting to a point where I could get therapy only to be continually stymied by multiple emails never responded to, multiple calls never answered or replied, and finally after one appointment for online therapy finally set up being ghosted by the practitioner.

I get that therapy is good, and I believe it can be helpful or else I would have never sought it out, but, I am amazed that anyone can get it.

I've also met people who have no money, no insurance, no job, but are able to get therapists with a single call, which is all the more infuriating because those same people have been consistently unable to help me do the same.

Money and tears and pleading won't open these doors for me and it almost feels personal, like the universe itself conspires to keep me specifically away from good mental health, so I have a nice reservoir of anger towards anyone who casually mentions that going to therapy will solve all your problems or that the conventional way of doing things is the right way to do them.

If human therapists make themselves unavailable to me both in person and online, then I'll make do without them.

Second counterargument: by practising, someone will eventually get good at it, and it becomes feasible to treat very niche or unprofitable disorders.

The quality of human practitioners is...variable, and they tend to focus on only a few areas.

but my meditation app didn't get bought by Google and shut down like I'd planned and I can only build so many to-do lists

I think llm could be helpful for therapy. I've never tried it for that purpose but it's the impression I get. Mostly it is learning to talk to yourself and chain of thought to a better mindset

Meh. I'll agree that chatbots shouldn't be allowed to claim they have credentials, like anyone else. But a lot of therapists do nothing except sit there and listen to you, tell you what you want to hear, give some standard advice you could have gotten from a self-help book, and hand you a bill. An LLM can replace the first three functions just fine.

If you can get your money's worth from talking to most therapists, you're probably self-aware enough to get it from bouncing your thoughts off an LLM.

> Maybe that sounds like a no-brainer: don’t trust Meta with your mental health.

Reply on news.ycombinator.com