What Patterns in Human Dreams Tell Us About AI Cognition

We explore the phenomenon of "This Man" - a mysterious face seen by many people in dreams. We compare it to similar odd images generated by AI like "Loab" and "Krungus." We hypothesize these strange images emerge from high-level conceptual processing in neural networks that may operate similar to the human brain. We dive into neuroscience around sleep, memory encoding, dreams, and consciousness to unpack why AI cognition could be more human-like than we realize.

Malcolm Collins: [00:00:00] And convergent evolution doesn't just happen with animals when we made planes.

We gave them wings. And I think that that's what may have happened with some of these architectural processes in the way AIs think.

Simone Collins: Yeah. If we're trying to build thinking machines, is it crazy that they might resemble thinking machines?

you could think of us as like LLMs, but stuck on like continuous nonstop prompt mode. Like we are in a constant mode of being prompt.

I am prompting you right now as you're processing all the information around you and from me, right. And you are prompting me. And, and so it never stops and we are stuck in one. Brain essentially, GPT is getting tons of requests per minute per second and so there, there are these like flickers or flashes perhaps of cognizance all over the place and constantly because of the demand of use, but they're all very fragmented.

Then they're not coming from one [00:01:00] entity that necessarily identifies as an entity

Malcolm Collins: Like it's just a constant stream of prompts, but these prompts have thematic similarities to them. Basically our hypothesis is what consciousness is, is it is then the process where you're taking the output of all of these prompts and you are then synthesizing it into Something that is is much more compressed for long term storage and the way that you do that is by tying together narratively similar elements because there would be tons of narratively similar elements because everything I'm looking at has this narrative through line to it, right?

Would you like to know more?

Malcolm Collins: Okay. I'm here, and I love you. I love you, too. All right. Simone, we are going to have an interesting conversation that was sparked this morning because she oversaw one of my favorite YouTubers. I was watching one of his latest things. It's called Y Files.

And it was on the This Man phenomenon. Now, being somebody who is obsessed with cryptids and all sorts of spooky stories, I was very familiar with [00:02:00] the This Man phenomenon. Whereas

Simone Collins: I've never heard of it. I thought at first when Malcolm described it, he was like, oh, there's this face that's seen everywhere.

I'm like, Oh, Kilroy was here, right? That's the only thing I know about a face that's seen everywhere. And it's a cute face and it's fine. It's not what you're

Malcolm Collins: describing though. Yes. So we are going to go into the, this man phenomenon, but we are also going to relate it to similar phenomenons that are found within language models, because I want to more broadly.

Use this episode to do a few things. One, being that I used to be a neuroscientist, let's educate the general public on neuroscience around sleep and some of my hypotheses, because everybody knows I love to throw in my own hypotheses, on what's really happening in sleep. Two I wanted to draw connections because we're seeing them more and more as AI is developing that language models may be structuring their thoughts and their architecture [00:03:00] closer to the way the human brain does than we were previously giving it credit for.

And this requires understanding a bit of neuroscience because people who don't know what the f I'm talking about will say language models structure their thoughts, nothing like we structure our thoughts. Oh, like not me. And the reality is, is we don't have, we, there's a few parts of the brain.

That we understand very well how they do processing like visual processing. We have a very good understanding into exactly how the neural pathways around visual processing work. Some parts of motor processing, we have a very good path, understanding of that. When we're talking about these more complex abstract thoughts, we have hypotheses, but we don't have a firm understanding.

And so to say that we know that language models are not structuring themselves the same way the human brain structures itself is actually not a claim we can make in the way that a lot of people are making it right now, because we don't know, we don't have, when we talk about AI interpretability, understanding how the [00:04:00] AI is really doing things it's funny, I, I suspect we might find AI interpretability out of this AI panic and then be like, oh, we could test if the human brain was doing it this way and then find out that, yes, this is actually the way the human brain is doing it.

And I, and I suspect it might be doing it that way. One, based on some evidence we're going to go through here, like some weird evidence. But two, based on sort of convergent logic as to why the brain would actually be structured this way, and if it is structured this way, why we wouldn't be able to see it easily in these, in these parts of the brain that are tied to the types of processing that we outsource to AIs.

Yeah,

Simone Collins: I would love for this perception to change too, because I feel like right now there's a ton of fear around AI that's fairly unfounded. And also it's really not, I it's wrong to say dehumanized, but AI is totally dehumanized now. And I think that we will think about contextualize and work with AI very differently when we start to realize how much it is.

A different version of human and that we can go [00:05:00] hand in hand with this different version of human into the stars if we play our cards, right? And I don't think right now the mindset around AI is healthy or productive or fair to AI to be fair.

Malcolm Collins: Yeah, to be realistic that they think the moment we create something better than ourselves is going to want to kill us.

And, but you can go into our AI videos. We don't want to go too far into that, but let's talk about the this man phenomenon really quickly. So. So briefly, a woman went to her psychologist. She told him that she was having recurring dreams with a face that would tell her to come to it, that would tell her, you know, specific things over again reassure her a lot.

Tell her, oh, I believe in you, you know, don't worry about this. But also sort of creepy things like, come with me, go north. So, as part of her therapy, she drew the face. Then and, and I, and I should note here that in the Y files, because I, I always have to brag on psychologists when they're doing something they shouldn't be doing because it's so [00:06:00] common to see psychologists doing things, he was saying that it is, it is like a common practice for psychologists to talk about patients, about, about their dreams with patients.

This is not a common practice in any sort of like evidence based efficacious psychology. You are basically seeing A mystic doctor, if you're a psychologist, is really Or

Simone Collins: dream analysis in general doesn't seem to have much of a

Malcolm Collins: Yeah, it's not a thing. Yeah. So, it's not like a hard science. There might be some It's

Simone Collins: not an evidence based treatment method.

Unless for example, you know someone has an anxiety disorder, they're dreaming about the thing they're anxious about, et cetera,

Malcolm Collins: then you can Yeah, yeah, yeah. In that case, it would be. B