Bad Therapy, Better Training, and the Promise of AI · Psychotherapy and Applied Psychology · Radar

AI and psychotherapy: benefits and risks1:57 – 7:00

1:57

And on to the first comment.

1:59

The first comment is about the episode that I did on AI and psychotherapy, although AI and psychotherapy has come up a few times since, with Dr. Tony Roumaniere and Dr. Alex Vaz, which I'll link to.

2:11

@Chris2455 wrote, "An LLM is always there when you are most vulnerable and desperately need help.

2:18

You don't need to wait a week or talk to someone.

2:20

It's instant support and it can be life-changing." So I had a few people write in with sort of similar positive sentiment like that.

2:27

I also had some people, not surprisingly, write in with some more negative sentiment.

2:31

@Tomorrow4Eva wrote, "This is what people ignore about AI.

2:36

They treat it like it's a rational entity.

2:38

It's not.

2:38

It's a predictive text generator.

2:40

Its job is to predict what text should come next.

2:43

It tries to tell you what you want to hear." So let me just give a few thoughts about this.

2:48

The first, let me hit on kind of some of the value of AI, and let me tell a little bit of a story.

2:53

A couple of months ago, my son hurt his thumb playing football, and if you're familiar with nine-year-olds, their ability to report what their experience is and how dramatic it is or how not dramatic it is or how problematic or not problematic it is, their self-report's not so reliable.

3:11

So you do your best to sort of-- and what I did, I did my best to sort of assess, okay, how bad is this? It's probably not so bad.

3:16

He hurts himself all the time playing sports.

3:19

But then he sort of-- it seemed like it wasn't getting better, where over the next day or two, it was still swollen, it was still bruised.

3:27

He was still complaining in ways that he doesn't normally.

3:31

So I took a picture of his thumb, threw it into ChatGPT, asked a couple of questions, and it was tremendously helpful.

3:39

And, you know, it gives that, "Hey, I'm not a...

3:41

this isn't professional medical advice," but it also said, "Yeah, this sounds like you should probably go to the doctor's and get it checked out." Next day, we go to the hospital, we get it X-rayed.

3:54

His thumb is broken.

3:55

It was a really useful tool.

3:58

I think for folks who are struggling with some mental health stuff, who are just like, "What am I experiencing? What's going on with me? Should I go and see somebody?" These AI chatbots can be a useful tool for giving folks just someone or something that has some insight or potentially some knowledge that could be useful for helping you make these decisions and figure out what's happening.

4:20

Of course, you have to take it with a huge grain of salt.

AI and psychotherapy: benefits and risks1:57 – 7:00

1:57

And on to the first comment.

1:59

The first comment is about the episode that I did on AI and psychotherapy, although AI and psychotherapy has come up a few times since, with Dr. Tony Roumaniere and Dr. Alex Vaz, which I'll link to.

2:11

@Chris2455 wrote, "An LLM is always there when you are most vulnerable and desperately need help.

2:18

You don't need to wait a week or talk to someone.

2:20

It's instant support and it can be life-changing." So I had a few people write in with sort of similar positive sentiment like that.

2:27

I also had some people, not surprisingly, write in with some more negative sentiment.

2:31

@Tomorrow4Eva wrote, "This is what people ignore about AI.

2:36

They treat it like it's a rational entity.

2:38

It's not.

2:38

It's a predictive text generator.

2:40

Its job is to predict what text should come next.

2:43

It tries to tell you what you want to hear." So let me just give a few thoughts about this.

2:48

The first, let me hit on kind of some of the value of AI, and let me tell a little bit of a story.

2:53

A couple of months ago, my son hurt his thumb playing football, and if you're familiar with nine-year-olds, their ability to report what their experience is and how dramatic it is or how not dramatic it is or how problematic or not problematic it is, their self-report's not so reliable.

3:11

So you do your best to sort of-- and what I did, I did my best to sort of assess, okay, how bad is this? It's probably not so bad.

3:16

He hurts himself all the time playing sports.

3:19

But then he sort of-- it seemed like it wasn't getting better, where over the next day or two, it was still swollen, it was still bruised.

3:27

He was still complaining in ways that he doesn't normally.

3:31

So I took a picture of his thumb, threw it into ChatGPT, asked a couple of questions, and it was tremendously helpful.

3:39

And, you know, it gives that, "Hey, I'm not a...

3:41

this isn't professional medical advice," but it also said, "Yeah, this sounds like you should probably go to the doctor's and get it checked out." Next day, we go to the hospital, we get it X-rayed.

3:54

His thumb is broken.

3:55

It was a really useful tool.

3:58

I think for folks who are struggling with some mental health stuff, who are just like, "What am I experiencing? What's going on with me? Should I go and see somebody?" These AI chatbots can be a useful tool for giving folks just someone or something that has some insight or potentially some knowledge that could be useful for helping you make these decisions and figure out what's happening.

4:20

Of course, you have to take it with a huge grain of salt.