AI Therapists Are Coming For Your Brain

Added: 2026-05-31

[00:00:00]And when we can communicate telepathically, does that eliminate all the problems that we have with leaders, with human beings governing people in corrupt ways?

[00:00:12]>> Now, to be clear, I think so people don't think I've lost my mind. Um, we're talking about like telepathic is like a neuralink like version.

[00:00:19]>> Yeah. Some version of that. Something that allows you to communicate without I mean, that's one of the things that Elon said to me when he was talking about Neurolink going to be able to talk without words. Yes.

[00:00:29]>> Like, >> oh boy.

[00:00:30]>> Yeah. Yeah. Yeah. Yeah. No, I think it's going to get >> and a universal language like something where you can communicate and we could really understand, oh, oh, we really are the same.

[00:00:39]>> Well, I would say again, but here's a human, here's a human values question, which is like, okay, if you are one of these people that has one of this thing, it's like, okay, well, how much of yourself do you want to expose to the world?

[00:00:47]>> Well, give you an example. Can the cops come get your Neuralink, right? Can Right. Can they come get your thoughts?

[00:00:53]Right. And so, you'll >> Isn't that a Dark Mirror episode? you uh pro probably you'll want to have yeah so you'll want you'll want to have again like American legal system you're going to want cops are going to need to get a warrant to get a transcript of your thoughts or maybe not maybe they can't get it at all because we decide that that's just a horrible road to go down in the American system we we hopefully will have some method for doing that you know in the >> unless the Democrats get in control >> in the Chinese system >> the CCP will come get it anytime they want right so so and again I just human values questions Yeah, we're gonna Yeah, we will be confronted with those questions. We will have to answer those questions.

[00:01:29]>> But I think >> the machines won't get us out of >> your perspective is ultimately it moves us into a much better place.

[00:01:35]>> I just we're going to we will be so much more capable. I mean just I mean it's it's almost a cliche now, but just like how about we start by curing all disease.

[00:01:42]>> Yeah.

[00:01:43]>> Like how about that? Just to get going.

[00:01:46]And you know look we still got work to do but like you know these things are like I said these things are already solving math puzzles that human mathematicians couldn't solve. They're going to start to do all kinds of things in biology. There's very exciting projects happening >> and maybe psychology as well like all the emotional issues that people have.

[00:01:59]>> For sure. Yeah. Like actually by the way there actually there there is there is actually there's one form of actual clinically provable therapy that actually works and it's called cognitive behavioral therapy. Um and it's 100% something that an AI could do. No question. Right. And so all of a sudden like might it make sense to have everybody have that? I don't know.

[00:02:18]Maybe. How do we feel about people having AI therapists? I don't know.

[00:02:21]Maybe we're going to think it's a terrible idea. Maybe 20 years from now, we're going to be wondering how do people function totally on their own without any help.

[00:02:27]>> Well, isn't there also an issue currently with like AI therapy gaslighting people?

[00:02:32]>> Well, it can and again Netflix scripts.

[00:02:34]So, so here's a problem that you you may have seen the industry's been dealing with, which is about a year ago there was a big problem that developed. So, there's this idea I think the philanthropic puts it is you want the uh you want the you want the AI to be honest, helpful, and harmless. Um um and and there's a whole bunch of questions in all three of those, right? which is like for example exactly how honest do you want it to be? Um right like do you really want it to tell you all the like all the truth about you know whatever.

[00:02:58]Anyway, there's that but there's also okay harmful okay well the harmful and helpful it's like okay do you want it to always agree with you?

[00:03:05]>> Okay well and then that that's what in the field is called the syphency issue.

[00:03:08]the AI is a sickopant, right? Sucks up to you, right? And so it's like, oh, I have a um you know, I'm I I um I I need I want to get a promotion at work. Can you help me do it 100%. You of all people definitely deserve this promotion. Um and then you go back the next day, I didn't get the other guy got it. That's so unfair. You were the person who really deserved it. Okay, so that's that's the easy version. The harder version is I have come up with a design for a you know, a perpetual motion machine. You have achieved a physics breakthrough that the greatest minds in physics have been unable to achieve. you are a singular talent and the fact that you haven't received a Nobel Prize, right?

[00:03:41]>> See where this goes. So, >> so that's feeding the that's that's that's taken the honest and harmless part like and helpful part to it's like too helpful. And so the the new models are backing off on that.

[00:03:51]>> So what I've done is I've gone the other way. I've I've you can load custom prompts into these things. And so I've loaded I've created a prompt and it basically says just give me the brutal truth. Just give me the brutal facts.

[00:03:59]Don't worry about my feelings. Just like immediately tell me the way that it is.

[00:04:02]And >> the thing just rips the [ __ ] out of me.

[00:04:04]like it >> and it literally is I actually think I have to change it because it starts every answer with here's why you're wrong.

[00:04:15]>> It's like this assumption is wrong, this assumption is wrong, that statement was wrong.

[00:04:18]>> Wow.

[00:04:19]>> You know, you really don't understand this at all. And then it like goes into detail from >> an education perspective though. That's amazing. It's amazing if you really want to grow.

[00:04:25]>> Exactly. 100%. If you want to grow and so so what do you what do you want?

[00:04:28]Probably you want something in the middle. Right. Right. But you gota Yeah.

[00:04:30]You got you got to you know human values question. You got to decide what you want. All right. Yeah.