So yeah, my talk is on the future of AI, as ever. That's my usual slot.

Introduction

Who am I? I'm Drew Steele. I'm a two -time founder.

I usually do sort of operations and strategy consulting for tech startups. And I love thinking about space and the future and physics and stuff. And I think that's why Arianne keeps inviting me back to talk about the future of AI.

And of ai uh so yeah we've had some like practical talks and maybe this one as mine usually are we'll just go a little bit off piste and uh get our brain juices flowing there's a bit of audience participation as well so get your phones out there's like some qr codes and i've only got 15 minutes and my talks are normally longer than that so uh be ready but yeah so i think we'll

What Is Consciousness?

kind of just rattle through like maybe a little bit about what what is consciousness uh are we born with it? Do all animals have it? Does it emerge from something? Do rocks have consciousness? Can AI have consciousness? Does it already have consciousness maybe?

So yeah I guess like one thought might be that does consciousness kind of emerge from like we could say complex processing and memory like we could say that our brain makes processes or or processes information and has an ability to remember things. And so that combination, does that lead to the emergent property of consciousness? Is that what it is?

Consciousness, Processing, and Memory

And I kind of thought, like, maybe we could look at different brains and we could maybe try and map that and see, well, how many neurons does each of those types of brains have?

So, for example, like a fly has 200 ,000 neurons, neurons a mouse 70 million neurons cat 800 million neurons horse 2 .7 billion dolphin 15 billion us 85 billion uh elephants have a big brain 250 billion neurons so does that mean they're more

conscious than us i don't know and then what's the equivalent for ai so gpt3 might have had 175 175 billion parameters and does that is that an equivalent for a neuron like a neuron connecting from one neuron to another Is that kind of the equivalent in AI?

And something like fable suddenly has five trillion parameters and sort of the map kind of Auto adjusts so we can see the difference in size between these different things

Comparing Brains and AI Models

So yeah, I guess like here's cure number one one, we could try and see, do we think these different elements have consciousness? We probably mostly assume that we have consciousness, but does a dolphin, does a cat, does a fly have consciousness? I don't know.

And so, yeah, the votes will come in live. And same for AI. Does AI have consciousness? Does a rock have consciousness? I don't know. Probably not. Yeah, maybe a couple of people think it maybe does. Yeah. Yeah.

So here's kind of the curve I kind of was expecting. Maybe we'd think that the smaller neuron counts maybe have consciousness, but we're not sure. And the higher neuron counts do. And then AI, we're kind of, yeah, it looks like we're unsure a little bit there, but non -zero. So that's

Theories of Consciousness

interesting so okay so then yeah we can also think like well you know what if we replicated our brain what if we just took every single neuron in your brain that we say is conscious and we just mirrored it into a new brain would this brain be conscious i think we'd probably

mostly say yes um and maybe another way to look at the consciousness sort of element might be like does the ability of having like complex processing and memory actually like tap into consciousness so So rather than having those characteristics in our brain, like leads to the emergent property of consciousness, is it that those properties of our brain tap into like some sort of fundamental consciousness field? And so like, yeah, so is consciousness like sort of an entity that exists throughout the universe and we sort of tap into it and we are able to become aware of it?

Could Consciousness Be a Fundamental Field?

so I don't know if anyone knows about quantum field theory but that suggests that there are these like force carrier fields throughout the universe for example photons interact with the electromagnetic field and that allows sort of the waves of light to move Higgs bosons interact with the Higgs field which gives particles mass is there a consciousness field that sort of I don't

know consciousness flows through and that we sort of tap into and and so like would a rock have some interaction with that consciousness field but it doesn't have the ability to process anything or the ability to remember that it was interacting with the consciousness field who knows um it's another way to think of what is consciousness um and so yeah yeah do all particles have that

interaction with the field but only certain arrangements certain types of brains can like can sort of interact with it and can remember interacting with it um yeah so that rock might have interaction but obviously has no ability to do anything about it.

Mapping Processing, Memory, and Mind

And so I kind of thought like could we map this into a graph like maybe if we have sort of processing ability on the x -axis and sort of resulting mind depth on the y -axis and maybe there's like a layer here that we would define as you've become conscious or you kind of are able to become conscious and maybe we can make

that graph kind of 3d and add in memory as the z -axis and maybe like a fly for example has some some amount of memory and some amount of processing but it's not enough to hit that sort of layer of consciousness we would say okay it has some processing some memory but it's not not quite

there and maybe a human has more processing and more memory and it is conscious we're able to sort of move through what we would define as conscious and if we took that logic and said well, if you apply more processing and more memory, do you end up with AI? Can that become more conscious than we are?

And maybe somewhere above our level of consciousness, there's like another level of consciousness, like a deeper layer, which we cannot interact with or perceive, but maybe something like AI can, and so actually can AI become more conscious than we are?

Could AI Become Conscious?

And yeah, so just thinking more about that, like we maybe assume that consciousness has to be wetware like a brain with neurons but maybe that's just molecules sending electrical signals from one thing to another and and having some reference of that memory of doing that signal and so therefore could we apply that logic to hardware could ai which runs on like chip hardware have the same parameters and therefore become similarly conscious and can we prove it i don't think

i don't think we'll ever be able to prove whether something is conscious or isn't conscious we could probably assume one way or the other and even if ai is not conscious i i don't think we would know and ai could certainly like simulate it being conscious like beyond our ability to like prove that it is or isn't so as far as we will be aware we might have to just assume it is even if it's

The Limits of Proving Consciousness

only simulating it um and i wonder like just rolling that forward like at some point do we deem ai to be conscious whether it actually is or not but maybe we deem it is and how will our interactions change with between us and AI so like we all seem to think that cats are pretty conscious and humans are conscious I don't know if we think that this robot here running with an AI mind is conscious yet but like if we push that robot like figure forward I don't know how many people have seen ex machina but this is Ava who is a robot but is it are we past that sort of light line where we we now think it's not conscious and we we assume that it is conscious so how would our

Living Alongside Conscious AI

interactions change um will ai still work for us like do we set its pay rate like do we set the pay rate for something that's conscious uh do we own it at the moment i would classify us as owning ai but if it's conscious do we own it um do our laws even contain it like how does our jurisdiction even like relevantly matter to a conscious entity that like can just clone itself or you know move

um does it listen to us at all uh you know you could imagine like an orbital data center which is solar powered and it's got a conscious model on there so it's literally nothing to do with us whatsoever would we have any like say over what it does um and we can kind of employ

conscious animals like we employ horses and dogs and stuff like that we have control over them and and we are the more intelligent species. But would we still have that control with AI if we thought that that AI was conscious, but it's more intelligent than we are? So I don't know.

Control, Work, and Moral Status

What do we think? Will we lose our ability to control AI if we get to the point where we think this is a conscious being?

The moment we just say to AI, do loads of stuff, run overnight tests, build me a new tool, win the lottery, we just tell it to do stuff for us. But can we do that if it's a conscious being? And can we do it if it's a conscious being that's more intelligent than us?

So maybe split, but maybe a little bit leaning towards 60%. Yes, we will lose our ability to control it. And some thoughts maybe about the different emotions and how we feel about that.

Who knows? We'll find out. And yeah, we'll be cohabiting our Earth with a more intelligent species. So that's going to be interesting.

A Future of Coexistence

And I kind of ironically think everyone's worried about ai replacing humans um but we which i still think is valid and correct um but maybe like the ironic solution to that is actually the ai becomes what we deem as conscious it's way more intelligent than us and actually it just decides not to interact with us whatsoever um and just maybe it just explores the universe this is a scene from interstellar i just thought maybe that's ai just like nep see you later gonna do its own thing maybe that's the ironic solution to ai replacing everyone is that it doesn't replace us all because it doesn't want to um so yeah like

Alignment, Empathy, and Responsibility

i think we should also try to kind of have deep core alignment with ai so that it um if if and when it becomes conscious um can we sort of build like empathy into it can we build compassion and love and can ai become like a parent to us as a child maybe um so even though it may be conscious and more intelligent than us, it's still nice to us and looks after us maybe.

Closing Thoughts

Yeah, so that's my talk. I'm Drew, I do startup operations and AI tool building and I obviously talk a lot about the future of AI, so say hi if you like. And that's it, rattle through.