0002: Consciousness

Hello this is Brandon. This is Ted. And we're going to talk about concerning AI. No, we're concerned with AI.

This is concerning AI. This is concerning AI. Today we're going to talk about consciousness. Yes.

And if our brains are too bad and we melt, you'll understand where the audio goes. I think, therefore I am. It's not what we're going to start. I don't know.

I just made me think of that when I said, I'm not ready. Yeah. My brain has heard all week trying to think about this question. I find it is harder than intelligent.

It is much harder than intelligence. Yeah. Does it make sense to start with a dictionary definition? And maybe that's not very adequate.

We can dispense with it right away. But at least the place to start. Sure. That's hard.

Memorize. The state of being awake and aware of one's surroundings. Awake and aware of one's surroundings. Yeah.

So like when I go to sleep, am I still conscious? It doesn't seem like it. It seems like I am unconscious. You're not awake.

You're not awake. And if you're like knocked out. Right. I'm unconscious.

Right. OK. That's not conscious. Yeah.

I'm unconscious. So awake and aware of one's surroundings. So is awake? Those two words are interesting.

Awake and aware. Aware is a tough word, right? Aware is a tough word. I definitely don't know with any of this consciousness talk, if I can assess your consciousness or not.

It's like, let me assess my own. No. Yeah, you can't know that I'm conscious. Yeah.

So if that is a supposition, we might want to examine that a little bit. But if that turns out to be true, then this whole conversation about AI and AI consciousness becomes much more difficult. Well, perhaps it has to do again. OK.

So coming back to intelligence, we ended up saying that the procedure is less important than the external correlates. Yes. What you can actually measure about what's happening. That's right.

That's right. So it's not looking at brain waves. It's looking at the action of that thing. The action of that thing takes.

Right. So we have to see the things that we said are intelligent, which is what do we say? Efficient, cross domain, future-shaking. We can see that.

We can see a person or a being or a dean or whatever. A thing. We can see it doing that. So what would it look like to see a thing being conscious?

Well, I have an example. So I encountered sentience and safety and, says two other words. And sentience is supposed to be feeling, the ability, and activity of feeling what's happening to you. And you're like pain, pleasure, that sort of thing.

So when a dog makes the sound, ar-ar-ar-ar-ar-ar-ar-ar, that sounds very clearly like a sentient sound. Like, that hurts. I don't like that. I don't want it.

And lots of animals would have similar. We could assess that they are exhibiting sentience. So if something a sentient is it to something very conscious, in the definition we had earlier, awake and aware of its surroundings. I think so.

It certainly needs to be awake. Or if it wasn't awake, we just woke it up. Like, poor water run it's head. It's not awake.

It's making that. If it's a dog and it's making that sound, it has to be aware of its surroundings because it's reacting to something that just happened to it. So when I feel pain, that means something to me. And when you feel pain, you react similarly to how I react.

But I have no idea whether you're perceiving it the same way. Same thing for the color red. Yeah, and this is an example that comes up in a lot of the different readings. This idea about color is that we can't tell what someone else's experiences with that color, even though it always comes back to a wave of light or a speck of wave lengths.

So again, I believe that you can, that somehow, there is some generalizable element of your experience of the color red and my experience of the color red, because functionally, we can talk about it and we can use it, our experience of the color red, to do useful things. Yeah. But I guess you could do that maybe without consciousness as well. Like a machine could perceive red apples and keep them on the conveyor belt and brown and do the apples, kick them off into the rubbish.

So that's actually perceiving red and acting on it for benefit without being conscious. Right. Or is that actually a rudimentary form of consciousness? I think that this is where the definitions fail.

It's a little bit because I can't even imagine that that could be considered conscious. That doesn't make sense. That's just, there's definitely a perceiving. It's like, it's just input output.

That's what computer programs are about. There's some input and there's an output. Well, that brings us to the big debate, or a big debate about consciousness. Is it just, is it deterministic or not?

Because if it is deterministic, then it's just a way more fancy version of stimulus response. You know, I think the apples are like, I put it into the pack these apples and ship them off to the supermarket. I see the brown yucky. I respond with, munch it in, you know, put it in the compost heap.

You say something kind to me. I respond with let's be friends. You say something mean to me. You know, I cut you out of my will.

I mean, this brings up the concept response, right? The conversation about free will, right? I mean, do we actually have free will. And do we need to solve the free will question to solve the consciousness question?

Can we separate those two? Well, can we come back to that for a second? Because I just had a thought. So I think that there are stimulus response processes operating in me that don't have what I would call the subjective harbingers of consciousness.

Such as, for example, my heart is beating. But I'm not thinking about it. And I'm not even aware of it unless I, you know, really add some extra instrumentation. You know, put my finger on my poles, use a stethoscope, an EKG, something like that.

But it's happening. And, you know, something is stimulating. In this case, my brain is stimulating it. So maybe that's not the best example.

A different example might be my hand touches something hot. I jerk it away before I even really perceive that it's hot. You know, before my consciousness perceives it as hot. Now, it definitely makes me think about babies.

I did a lot of things about babies this week. And trying to decide whether babies are conscious or not. Because a baby, you know, you put something near the baby and the baby moves its face to try to stock up that thing. Like that seems like instinctual, just like the DNA talking.

It doesn't seem like there is any decision-making or necessarily awareness. It's just the basic stimulus response. Right, again. And on one hand, it seems like a baby, even though they're not talking yet and not really interacting with the world very much.

I think they should be. I think I want to say that they are conscious. And yet none of the definitions help to me figure that out. You know, it's funny because watching my guys go from being babies to not was sort of the process of watching something wake up.

Yeah. So it could be that there is a slow gradual process that starts in utero where the baby can hear its parents talking and that there is an acquisition of consciousness that's slow. And it seems like at the point that they can talk, then we know there's something else going on. But that process doesn't just start when they begin talking.

But it made me think about the Helen Keller story and how when she talked later about her acquisition of language, she talked about not being able to have any thoughts before that moment that she got the word water. You know, it was written on her hand that her teacher helped her with. So she had sentions. Like, you could definitely perceive pain.

And a baby can definitely perceive pain. But not consciousness in the sense of being able to put together the idea, I am, you are, that is. Right. Right.

Because she was missing a couple senses that most of us have, she had a different kind of experience. But babies have those things. Babies have those perceptions. They're not really able to use them very well at first.

Well, they react to their parents' voices right away, don't they? From the time they're born. They start developing that pattern recognition of faces right away. And that's one of those things.

Quickly, quickly. Yeah. I mean, language to me feels like such a big deal. And yet, this consciousness cannot be a function of language.

It seems like there are so many examples, counter examples. But now, what about the chimpanzee that learned some language, learned some sign language? Or a gorilla? Were they conscious before that?

Were they conscious before that? And then they were afterwards? I find the whole conversation very confusing, to be honest. Well, how about we take it from a different perspective?

Like looking at, are there higher states of consciousness than the typical human consciousness that we've had since we could first remember? For example, people talk about perceiving all of the universe as one integrated whole, as a type of elevated consciousness. Another thing that's less woo-woo for some of our listeners probably is, if you have calculus, this is another language, if you have calculus, then understanding what's happening when a car jerks in terms of the second derivative of velocity. Or it's maybe not the most perfect example.

But it does seem like having calculus creates a consciousness of certain phenomenon and awareness of what's happening, this is happening. So you say property A is happening because now you're able to perceive it as happening because you have the tools to do that. Is this helpful? Is this going anywhere?

I don't know. I did a lot of thinking this week about distinctions. And there are some distinctions that may or may not be sort of provably true, but they might be useful anyway. The distinction calculus seems like it's useful to understand a phenomenon that's happening.

Calculus is actually provable, like this is a thing. But it makes me think about how we talk about the sunrise. When we say the sunrises, we are misperceiving what's actually happening in the world. The earth is actually spinning.

The sun isn't rising at all. It's the earth is spinning and therefore we are now singing. I don't know what the new language is. We actually describe that reality.

But it reminds me of a story about what's the guy with the domes, the geodesic domes. What's that guy's name? I forget his name. Buck, Mr.

Fuller. Buck, Mr. Fuller created a language about up and down. He said I'm going out and I'm going in thinking about the earth.

When he went upstairs, he was going away from the center of the earth. So he was saying I'm going out and when he's going downstairs he's going in. So to create in himself a new sensation of what was true. Because there really is no up and down except for my own point of reference.

He was trying to create a different point of reference. The point of reference of the earth as a whole. So to use the term conscious there, in using that language he was conscious of the earth being a spinning ball. Yes.

When you use that language and when you use up and down the consciousness was conscious of the earth being a flat plane. That's right. Which energy of your life mostly doesn't matter. Okay so this brings me back to the weight but why, who am I?

I'm a huge fan boy of. He has a great essay about time and it shows all these different sort of zooming and doing zooming out, zooming out from a moment all the way to the history of the universe. And in one of the images he shows how little of the time that anatomically modern humans have been conscious. No.

Yeah. As in how they have language. Right. So it's not, I think that the assumption there is that the brain and the actual hardware, not a huge difference.

But with language suddenly there is a huge difference that comes. Okay. So even in writing it takes that to another level. So even without agreeing on a definition of what consciousness is, I think what we're maybe getting to is that we could say that before language and after language is a state change or is a quantum leap in terms of a level of consciousness.

Is that, is that, is that some reasonable? Yeah. Yeah. So I think, I think we have to say consciousness is not a binary property.

Why? I agree. I agree. I agree.

I agree. I agree. Even though we don't, even though we don't yet know what it is or even have a good way to talk about it. It does not appear binary.

It does appear that there is some continuum. So the definitions or short hands for consciousness might be synonymous with self-conscious. Because for example, I've read about studies where, you know, if you mark Greece on the side of a dolphin, you know, it goes and looks at itself in the mirror is demonstrating self-consciousness versus I'm not getting this right. No, but I think, I actually think that this is really interesting.

The awareness that I am itself. I am. I am. It's sort of this, well, it's similar to I think they're for I am, but it's the other way around.

It's I am. Therefore, I am. But I am. I am conscious.

There's something along those lines. Yeah. I mean, you know, developmental psychologists talk about when babies go from perceiving mom and dad and everything else as sort of part of them as one undifferentiated whole to having a sense of self and other. And that's a pretty big developmental milestone.

But I don't think it's necessarily clear that all animals, all mammals have. Right. Well, I mean, the dog that barks at the mirror thinking it's another dog is the opposite of this, right? Is this not seeing itself as a self?

Maybe. Or maybe not. Maybe just not recognizing that that image is the image of me. But I don't know.

Maybe that's something different. So let's try to go back. Let's step back and try to operationalize the word consciousness as a as having some utility to future discussions of concerning AI. Yeah, I'm not sure that it is.

But let's try it. Let's try it. Well, is there a distinction that matters here? Could you have a an intelligent, a generally intelligent machine that had intelligence at the same level as humans that was not conscious?

Generally intelligent at the level of humans, the way that I think about that question is that it has to have everything that humans have, including a feeling of a body, including emotions. And it seems like a sense of self would have to be baked in there because human consciousness. Well, let's come back to the intelligence definition again, right? So cross domain future steering.

Efficient. Yeah. And then the efficiency is how sort of interesting particular intelligences that are at the same level are. So cross domain future steering does that require feeling pain, the ability to feel pain?

I think no. I would say it doesn't. Yeah, you know, I think for example, even among humans, there are people who don't really feel the pain of others in the same way, empathy and mirror neurons and that sort of thing, right? But are very good at future steering.

Yep. I agree. And so coming back, I disagree with the assertion that it has to have everything that we have in order to be as at the same order of intelligence, general intelligence that we have. And I think consciousness may actually be one of those things that doesn't have to have.

It maybe doesn't have to be aware that it's a thing and that we're, I guess it has to be aware that we're things. So what's that difference, you know? It doesn't have to be aware that we're well. Well, it is being aware the same as sentience.

Like part of my answer came from something that I read, Rick Kurzweil, I'll say it a long time ago, probably more than 10 years ago. He said, when the machines say I am human, we will believe them because they will have everything that we have. They will have the same experiences as us. They will have the same perceptions as us, the other same kind of thinking that we do, even the same spirituality or things that we don't understand where that comes from or why that's there.

And so I wasn't really answering your question, your question was a different question that I answered from that perspective of something I heard said a long time ago. So when you said can we have a machine that meets our definition of intelligence from last time, cross domain, future theory, I think the answer is can we have that without having to feel pain? I think the answer is definitely yes. That machine will not tell us that it's human because it won't have that experience of itself.

So yeah, I mean, I think that Kurzweil's definition there or assumption there is crazy. It's generalizing from one in instance of general intelligence and saying that all instances of general intelligence are going to be in this tiny little area in intelligence space. What if spiders had evolved to be intelligent, generally intelligent? Yeah, I think that you would be dealing with a vastly different intelligence than human intelligence, but it could be just as instrumentally effective.

And the goals might be vastly different, but it could be equally good at achieving those goals and steering the future. Yeah, I mean, there's bigger questions here. Like why would AI be more like human than it would like spiders? Because it's creation?

What's based on humans? I think this is fascinating because I think this gets right at the core of why Kurzweil is excited about it and thinks there's nothing to be worried about because he can only conceive of human-like intelligences. And the people who are worried can conceive of all of these like, I would be totally freaked out by a spider that is as intelligent or more intelligent than me. That would not be good for me, I don't think.

No, definitely not. But I don't think when Kurzweil thinks of these machines, he's thinking about it all. He's thinking me better. Yeah, I think that's right.

Which I think that leads us to a future topic, the topic of skeptics, the topic that there are people out there in the world who don't think that there's any reason to be concerned. And that is a topic that we will come back to on the next episode. Yeah, let's do that. Let's talk about the people who think that maybe this podcast should be called Not Concerning AI.

Right. Before we go though, I just would love to see if we could sum up, did we learn anything about consciousness? Are we any closer to having a working definition? Does consciousness matter?

We need to talk about the Neodegrast Tyson idea that maybe consciousness is asking the wrong question. I don't like that idea. Will you unpack that for us a little bit, Ted? This example of a pre-section.

Pre-section or whatever. Yeah. So for decades, we did not understand the precession of victory. And we did understand why we were seeing what we were seeing.

And we made up a planet, Volcan, and said that's going to explain, you know, of course, that's not right. And it turns out that the answer is general relativity, that we wasn't precession at all. We just, our perception was seeing it that way. And so we didn't need to ask a question about why is working pre-processing because it's not.

Yeah, right. That's not even a question anymore. That's right. So maybe the same thing is true with consciousness.

We have examined this as humanity. We've examined the question of consciousness for millennia. And maybe for our purposes, it's not a useful question. All right.

Well, let's just go with that for now and we can come back to it if we need to. And next week, let's talk about skeptics. All right. So for concerning AI, this is Ted.

And this is Brandon. Signing off.