Hello I’m psychologist and today I had an idea of making AI psychologist. Cognitive behavioral therapy CBT is very structured knowledge and guidance is very simple.It can improve while being used as well.This idea gave me spark and I started to do research with who I can share this.AI psychologist can help so many people to improve their Mental health.
Any suggestions, comments advices can be helpful to me.
1 Like
My first comment would be that I doubt OpenAI would allow someone to do this just via the API. I think it would require a direct partnership with them to even begin to pursue it.
I say this not because I think it’s a bad idea—in fact I think it’s a great idea—but because the risk/reward ratio simply doesn’t make sense at this point.
The models are still to prone to hallucination and confabulation and don’t have anywhere near the capacity to stay properly aligned for something as critical as working someone through their mental health conditions.
Someday, absolutely, yes!
But, we’re still in the infancy stage of large language models. Change is happening very fast, so in a year my concerns might seem quaint, but I genuinely feel we are at least two generations and one major new idea away from an AI capable of safely taking on this role.
It’s certainly worth exploring though. If there’s a good metric by which you can quantitative score a therapist and can show a GPT-based therapist performs on-par with human therapists, then you can democratize access to mental-health services and that would be a monumental net good for humanity.
1 Like
Thank you so much for your comment.I believe in AI without hallucinations.And actually Human can have it too.
I believe AI can work with Human more faster.Although there is somethings like human human connection or Compassion that will be ‘‘out’’ of AI. On the other hand a lot of people want to work with their thinking and are ‘‘she’’ or don’t want to SEE anyone and this is perfect with them.Anyways I believe AI psychologist can help millions of people and make more healthy Humanity and more in JOY.
Completely. It can be not only beneficial and lead us to new thoughts, it can not only act as an adhesive to people who have scattered thoughts (such as myself), but it’s also incredibly therapeutic to simply have something to speak to, and work with their thinking.
I believe GPT naturally is therapeutic, but trying to model it as a psychologist may put some people off and do the opposite of what you’re intending.
1 Like
General GPT is not very cooperative in this matter. I did not study directly in therapy. But I have an interest in psychology. and studied in the area of consumer behavior which emphasizes control rather than treatment It may seem scary, and it really is. If there is no law to control it, marketers may use hypnotism to their customers.
As for AI, I find it both positive and negative. Negative impacts are likely to occur in communications, education, criminology, and possibly beyond. But AI is not the cause of all problems. Some things come from the images we once created. It is also the same problem that has arisen and remains the same as before. AI may just be a new culprit, but no matter what, we will still benefit greatly from its use, such as being able to do a good job of creating quality knowledge. Be able to provide appropriate knowledge for each individual.
What is of concern is the regulation of its use in illegal crimes. I feel like I might not respond to my criminal identity from now on. from technological leaps Those who commit crimes are associated with development. And it will be easier to add because the former will be a group of technological offenders who do not necessarily have to be people with the body of a violent offender. In addition, technology until reaching AI will be related to groupings related to Occupational behavior (I’m not sure of the specific term) will start with a group that engages in delinquent behavior when given the opportunity from time to time. and develop it into a habit-profession-organization
As for prevention or correction methods, they are often difficult because they do not need to be directly used to commit an offense.
I still lack a lot of knowledge and information that will lead to a solution, but just one science is not enough.