Levin

 

An interesting question

49:23

might be: is there anything in this universe that is zero on the cognitive

49:30

scale? Because I think very simple things already are not zero. This is a hard question. I'll tell you what I think about it, let's ask what

49:44

would the most simple, the most basil, the most basic versions of agency

49:50

look like, what do you need for that? People say: you think that

49:55

the rocks have hopes and dreams like us? No, you have to scale down, the point isn't that it's going to have our level

50:01

of cognition. What does the most minimal level look like, the smallest possible level? I think you need two

50:06

things for that, you need some degree of goal directedness and that, in William

50:12

James's definition, is the ability to reach the same state by using different means. So you need some degree of goal directedness and your actions need to be not completely explainable by current local conditions. If

50:30

what you're going to do is completely determined by all the physical forces acting on you right now then you're

50:35

probably some sort of billiard ball. Those two things I think even particles have

50:42

those because least action type laws in physics tell you that there's goal directness baked into the

50:48

the bottom levels of the universe and quantum indeterminacy gives you a really dumb version of not

50:54

being predictable by local conditions. It's not great cognition because it's random, but it's something and so here's what I would say: if there is any kind of definition to life I think we call 'life' those

51:07

things that are really good at scaling those up, so if you've got a rock it has no more capabilities than the parts that

51:13

make it up, it didn't scale, it's sort of lateral, it's bigger but that's it. If you've got something that's alive it's

51:18

cranking up the agency, the indeterminacy and the goal directedness across every scale of organization and

51:25

the cognitive light cone is increasing as a function of time. My friend and colleague

51:33

Chris Fields who was a brilliant physicist, among other things, I asked him: can you have a universe in which

51:40

there is no least action, in which there would be zero and he said that only in a universe where

51:46

nothing ever happened, in a completely static universe where nothing ever happened it would be zero but as

51:52

soon as you get interactions already you know you've got the basics of least action. So I believe

51:58

that in this universe there is no zero, I think everything has some capacity but some things scale it up

52:04

in an interesting way and some don't and that means that for some things the tools of behavioral science are going

52:10

to be applicable and for some things they're not. That's why this is not the same thing as ancient animism

52:16

where you saw a spirit in every rock because it actually does not pay off to treat rocks with tools from behavioral

52:23

science but it does pay off to treat gene regulatory networks that way and it absolutely pays off to treat various

52:29

animals that way because that's how humans train dogs and horses knowing zero neuroscience because there's this

52:35

amazing interface that some animals expose that allows you to train them. Then

52:40

the last thing you talked about is this inner perspective, I try

52:47

not to say too much yet about consciousness per se but we can say something about inner perspective and which systems have it

52:54

and which systems don't. Here's what I think: imagine a landscape, a kind of a hilly up and down landscape and you've got a bowling ball on this landscape. In order to know what

53:04

this bowling ball is going to do and to make it do whatever it is that you want it to do all you have to do as an

53:10

external observer is to pay attention to your view of the landscape, your

53:15

third person view of that landscape tells the whole story. You know where the hills and valleys are that's all you have to know, you don't need to take into

53:21

account anything else. Now imagine a mouse on that landscape, when you got a mouse on that landscape your view of that

53:28

landscape is not really that relevant, what's really relevant is the mouse's view of that landscape because he might have been rewarded and punished at different times in different locations, he's seeing that landscape in a completely different way. There's a

53:40

a valiance map sort of superimposed on it. He's got his own opinions about where he's going to go and so for all these

53:47

different systems you can ask to what degree (and, again, it's not going to be a binary thing, it's never going to be

53:52

a binary thing) but you can ask to what degree do I need to take the perspective of that system and ask what does it see

53:58

and what does it think about what it's going to do? To the extent that you have to do that a lot you're dealing with a high agency system that has an

54:05

inner perspective. To the degree that you can get everything done with your own model of what's going on then probably

54:10

it doesn't. So that I think is a heuristic by which we can start to say how much inner

54:18

perspective can I expect from these things. I am not claiming that today's AI architectures are mimicking human brains. I don't think they are. I

54:30

don't think they're mimicking human cognition although they do have some interesting things in common that people don't realize. The point about AI is not that you have to be a human for us to have to be kind to you, there are many living beings

54:44

that are nowhere near humans. The same thing for danger, you don't need to be human level or above to

54:49

be dangerous, there many really dumb things that are extremely dangerous so the thing we have to understand is that

54:55

you cannot pin your hopes for being able to distinguish moral worth and things like that - you cannot pin that

55:01

on what you are made of because there's nothing magical about a protoplasm, you cannot pin it on being an evolved

55:08

biological being because there's nothing magical about the random meanderings of the evolutionary process which doesn't optimize for anything that we care about, it doesn't optimize for intelligence or meaning or

55:20

anything like that, it optimizes for survival for copy number and none of those

55:25

things are really reliable. You can't say just by knowing what

55:32

something is made of or how it got here. We have to have principled frameworks for making that decision and we

55:37

don't have them yet in biology very much.