Skip to content
0045: We Enjoy Our Stories episode artwork

Aug 8, 2017

0045: We Enjoy Our Stories

from Concerning AI | Existential Risk From Artificial Intelligence Hosted by admin

Is all AI-involved science fiction the same? Concerning AI | Existential Risk From Artificial Intelligence is made with Jetpack Podcast. Full show notes and every episode at https://concerning.ai/

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

0045: We Enjoy Our Stories

of matches

Transcript Auto-generated, may contain errors

Welcome to Concerning AI, this is Ted. I think you mean to start with good morning, this is Concerning AI. Good morning Vietnam. Yeah, AI-NOM.

Yeah. Hopefully, there's not the same sort of screw up associated with AI that we historically associate with. Well, actually, I guess it depends on who you are, how you look back at Vietnam, right? Oh boy.

I don't want to go there. I know, but it's a good lead in. I'm trying to lead into our topic today, Ted. Yeah.

Yeah. It's about bias. Okay. So, you know.

What about bias? Well, there's several different ways you can tell that story. And I think there's several different ways you can tell the future, the near future story of AI and other things. And I wonder why you and I are drawn to the particular sorts of stories that we embrace.

I love the stories we read. I just want to eat them up. Yeah. I just stayed up late finishing another one of these singularity recipe books.

I think you said the word recipe, right? The AI fiction recipe. Yeah. It's not just the AI fiction recipe.

It's kind of like our AI fiction recipe. Yeah. Should we talk about what you meant? We let people know so they can slurp this stuff up too if they want to.

Okay. So, what I think I should do is I should describe in generic terms kind of like what happens in the book a little bit and then I wonder if people can guess what the book is. And my hypothesis is they won't be able to because they all follow the same formula, the same recipe. We'll have lots of different guesses.

Okay. Cool. All right. So, here's the recipe.

Super smart inventor type person male course invents world changing things in an Elon Musk sort of mode. The lone hero that comes up with stuff that's kind of Einstein for our age or Newton for our age. And there is a character that's sort of like your protagonist from a little bit from afar. They're just so amazing.

You can't really get inside their minds. There are people around them that you ride along with. And one of those people is the special ops person whose combat ability is amazing. And then there's the hot chick who is beautiful and funny and smart and falls for the brainy people.

And the protagonists are careful but not careful enough. And they learn that it's just through some sort of luck that they manage to avoid destroying humanity. And there are the anti artificial general intelligence people and let's move forward at any cost. And it's great if we go extinct because we are bad people, groups and they're all struggling.

And in the end, we find our way through to, as I'm saying this, I'm realizing that they aren't all exactly alike. But they're do seem to be some themes that come through. Like this genius person. Have you noticed that as well?

Yeah. You think it's wish fulfillment? Do you think that you and I kind of want the world to be a place in which there are these individual geniuses that tower intellectually about the rest of us or that we could be that person maybe? I don't know.

It's an interesting speculation. If we are not in the mainstream in our love of this kind of fiction. Well, that's a good question because I think actually that's maybe part of what the issue is. I think it is kind of mainstream-ish.

It could be mainstream and more appealing to people like us also. It could be both. It could be that our DNA, we've speculated that the value function for human beings could be this alpha, be well-liked, be accepted and have high-wuffy. That's what we think our DNA is.

The result of evolution has been that our DNA has that kind of baked in. That may be- And it may be that in our group the way to get to be there and the status is available through intellectual exploits and invention and engineering and those sorts of things. Well, yeah. I mean, throughout history, Braun has been more important than brains maybe.

Well, so that's why you got to have the special ops person in there so that you can kind of tap that as well. Yeah. And if you can do that through technology, that's even better because our bodies aren't like that. Our bodies are more attuned to the intellectual pursuits than the physical ones maybe.

I don't know. Well, you know, it's a little bit funny. I was just talking to Eric yesterday. I'll bring him up a lot on the show.

He's a good friend of the show and a good friend of us. And he was talking about- I think I mentioned one of the essays that the same author had written that I was just reading about how recent advances in technology are making it really hard, are destroying science making it really hard to write science fiction because you can't see past the conclusion that humanity is not going to be recognizable in the future of 100 years or 200 years from now. And Eric made the comment that he writes all these stories about these trees that have got kind of some collective consciousness or whatever. And he'll read it to his partner and his son and he'll finish and he'll be like, when's the protagonist going to show up?

And so yeah, I think there may be truth to that as well. Like if you try to write about anything that's some sort of collective consciousness or something that's not anthropomorphic, how do people connect with that? Yeah, okay. So I was just talking with my spouse yesterday about a book that we just read that is maybe speculative fiction, not AI based.

I don't know that the author would call it speculative fiction or not. She writes speculative fiction. And then this book happened to be with a first person, first person point of view, a person with autism in the near future where the world mostly looks the same as today, but they have some advances in biotechnology and they have some cures for autism, at least for babies. But then the protagonist was too old to have that treatment and he's already an adult and the details don't matter.

But what I was thinking about how unsatisfying the conclusion of that story was. And I was thinking, well, what makes me satisfied with speculative fiction, with future thoughts? And I was thinking, well, good science fiction introduces a world and it can be very hand-wavy at the beginning when you're introducing the world. But then you just live out of story in that world and you're not hand-wavy at that point.

You just live within the rules that you've set up for yourself. The reason I felt like this book didn't work very well is because the hand-waving has came in the conclusion. And it's very dis- because there's nothing recognizable human in that world. And so does the future have anything recognizable human in it?

All right. So then I guess that raises the question. Is for our listeners, maybe have you read anything or seen any fiction or read any fiction where protagonists are not anthropomorphic, not recognizable human? Can we do world building that leads us in a different direction?

Well, one of the questions that this makes me ask about myself, well, okay, so let me go back for a little history for myself, first of all. I read a lot of Hardy Boys books as a kid, which if you look at those, those are just super clear formula. They're very much wish fulfillment kind of books. And as I look back to the time period when I read those books, I feel like it tells me a bit about myself more than it tells me about the world.

So now I come to this conclusion that there's a certain recipe for the AI fiction and other things that I read that maybe it actually tells me more about myself than it tells me about the future or about the worst day of the world. And I feel like there may be some sort of systemic bias that is there that I could possibly do something about if I were able to characterize it better, if I were able to see it, maybe. What do you think? Do you think that the fiction and stuff we read is actually probably telling us more about ourselves than it is about the future of the world?

Well, yeah, I think that speculative fiction has always been more about the present state of humanity than about the future anyway. I think that fiction written in the 1940s tells us more about the 1940s than about the future that it was predicted. Oh, isn't that so true? Yeah.

Yeah. Look at Orwell's 1984 written in 1948. I think it tells us more about his time than about any projected, you know, the actual 1984 or any future. I mean, we certainly could read into it today what we think about our current state of affairs, but that's a different thing.

I think that authors are always trying to tell something about their own time. So that kind of brings up one of the posts in the Facebook group. IBM is clueless about AI risks. And there's David Kinney, IBM senior vice president for Watson and Cloud, told the US Congress that Americans have nothing to fear from artificial intelligence.

Blah, blah, blah. And then, you know, there's Mark Zuckerberg, who's like, you know, I wish that Elon Musk and Stephen Hawking and these others would stop scaring people. There's really nothing to worry about. And I had this moment in my bathroom.

Don't worry. This is very kosher for a talk. Anyway, I had this moment in my bathroom where I turned on the water and was looking at the faucet on my sink. And there's a drop here, you know, a little like imperfection in the metal there.

And this had this overwhelming rejection of the simulation hypothesis for a moment. Like, how could you possibly simulate at this level of fidelity, the world? And if you reject the simulation hypothesis, you know, what conclusion does that lead you to? I think it leads you to that you're in base reality and you have to deal with the, you know, the, the Fermi Paradox, which I think argues probably for there's a great filter ahead of us.

A great filter that potentially prevents the development of AGI? I don't know. So it's quiet out there. I mean, there are lots of answers to the Fermi Paradox, of course.

So I feel like I'm spinning. I guess it's probably is kind of obvious from my, what I'm saying here, but you know, I jump from one thing to another. I'm like looking for something to grab onto to understand our current situation and what might happen over the next few decades. And the conclusions are just so dramatically different.

I guess that's why we call it the singularity. Yeah, like it's all. We can't see beyond the event horizon. Yeah.

And if you, if you take any one of these things and really like chew on it, like the simulation hypothesis I just brought up again, it's, it's, it's consequences are so huge, you know, that if the simulation hypothesis is true, then we probably already have AGI and we made it through okay, just to be able to simulate at this level of fidelity, the computational resources available have to be huge, huge, right? So in that case, we have an existence proof that navigating through the scenario of AGI is as it has happened at least once. I guess that, yeah, it raises the question of can the simulation at this level of fidelity happen without AGI, you know, with just massive computation resources? And something would have to happen for that to occur, I think, like, you know, there'd have to be some development that I can't imagine that would, yeah, I just, I can't, I can't believe that.

And maybe AGI is the wrong term. Maybe it's just super intelligence. Right. You know, I mean, that's, I guess that's when I say AGI, I immediately go to some sort of artificial super intelligence.

And by artificial, I just mean it's not recognizably us as we are right now. Yeah. Yeah. Yeah, I think the book, Nexus was, the series, Nexus was nice for that because the, the super intelligence was a human who had uploaded and then had a massive, massive increase in her level of intelligence because she had a lot more competition available to her.

You could say that she was still human, although she no longer had a body. So I have to tell you the one thing that came out of this last book I read, I'm going to tell you what it was. It was, it was, uh, art, it was called Infinity Born by, uh, David Simpson, maybe? I can't remember exactly who the author was.

I think that's who it was. And the reason it relates to the, to the comment about the, the upload and that person is that the one thing that really struck me in this book that I hadn't really considered very carefully was, um, in the book, an uploaded person is in a body that has, it's like a human body, but it has a computer, an optical computer for the brain instead of, you know, our gooey gray wetware. And the clock speed had been artificially, um, damped to operate at human time scales, but this person gets into a situation of duress and they have to let the computer go at its natural speed. They don't know how to, they don't know how to go between, they don't know how to sort of speed it up by floor acts.

They either do it at the speed that they're holding back to or it goes full speed and full speed is a hundred thousand times as fast as a normal human brain operates. So it is a human brain, but it's running a hundred thousand times as fast. And the consequence of that is just awful. Like we really went into that a lot.

Like how, you know, between each keystroke might be a day as they're typing a message and they're running, Pell, Mel, six times as fast as a human being can go through this, this, uh, this building to get to the bad soldier person that they have to deal with. You know, it's, it's years of subjective experience where all they can see is, is this super, super slowly changing fixed picture. I mean, it does change a little bit, but if every second is 48 hours or whatever, you can imagine how the picture wouldn't change very quick. There's just no external stimuli.

There's nothing you can do. You're just trapped. So, so this comes back to shoe. I think, uh, the, the woman who was uploaded into the quantum computer in the Nexus, children becomes the first post human.

How can you not go crazy if your brain is so much faster than reality? Yeah, the way, the way they handled that in her was that the AI had lots of different relationships and conversations happening at the same time. And then ultimately found that it wasn't enough and she needed to go off with the other AIs and disappear from human world. That's how she solved it.

But if you, if you, if you are somehow, if you're uploaded, AI has to live in a human body and interact with only one thing at a time, or if, even if you're trying to globally manipulate the world to get what you want, which is what she was trying to do, it's tough. Yeah, that, that, exploring that scenario in the book, the way that it did, um, just impressed me. Okay. Uh, impressed me how level grass of the situation I really have.

There's, there's all these hugely profound in it. The, the consequences are far reaching and these, these edge cases, these sort of scenarios I've never considered that I feel, I feel frightened. I just don't have a handle on it. Yeah.

Well, yeah. And for fiction to really work, the, they need to change as little as possible from today, which makes it by definition, not realistic. Instead of having all the different exponentials happening, they choose one or two or three, right? Where some things are changing, but mostly things are recognizable to us.

That's the other reason why the collective intelligence story about the trees is hard, because it's too different from our game of reality. Yeah. Old, um, Azimov stories, you know, with the three rules of robotics, they had the positronic brain, right? The world was mostly like it was when he was writing it, but then they had this positronic brain and it was, you know, it was there were these robots and that, you know, that was all the change.

And the positronic brain was like human two X maybe or something. Yeah. Not human thousand X. Yeah.

Even human thousand X, right? And you just said a hundred thousand, I was thinking, one hundred thousand doesn't seem like enough. That doesn't seem realistic. It seems like the exponential takes over and all of a sudden it's not a hundred thousand anymore.

It's, you know, it's a million. It's a hundred billion. It's a, you know, it seems like that moment in time when it's a thousand X would be so brief that it doesn't, it wouldn't be worth writing a story about. Yeah.

I think the hundred thousand X comes from this David Simpson calculating or starting from the premise of how fast signals signals propagate in the human brain versus the speed of light. And so if you had a computer, an optical brain that was about the same size as a human brain, but the signals propagate at the speed of light instead of 50 to 200 miles an hour, then that's when you get the 200,000 X speed up. Yeah. And David Simpson has used that before, that idea of, of that kind of brain.

I sort of think that's a little bit like the positronic brain. There's a certain way that he has chosen to make the brain so that you limit things because once you connect it to the cloud, then I mean, you still have the speed of light problem, but you can offload, you know, a bunch of computation. What does that mean though, offload computation? I don't know.

I mean, I don't know. If you have the architecture of the human brain, then all you can really do is shrink it and make things move faster. You can't like spreading it out in the cloud, it actually slow everything down. You know what I mean?

Like, like if you take a sequential algorithm running on a single computer and you split it up and now parts that have to communicate across different computers across a slow link, then you got to really be doing each of the parts way, way, way faster because now you got the slow thing. But that's not what would be happening here. What would be happening here is that you're taking something that could operate. Yeah.

I just, I don't see how it could speed up the human brain to spread out. I mean, I think it's because we don't understand parallelism. I think that's where it has to be parallelized. So some of these things have to be parallelizable.

Otherwise you're absolutely right. If it's serial, yes, it won't help at all. Well, but that's what I'm saying too, is that the human brain is super parallel. And so if you have that architecture already, the way to make it as fast as possible is to make it as small as possible with the slowest, you know, to decrease the slowest link from doing human, which is also, you know, my hope for what might happen with super intelligence that it just doesn't need all the resources because it's just so slow to get from one place to another and wants to be tiny and dense and fast.

Yeah. Am I raining on something here? I feel like I'm being knelly negative. No, no, I mean, I think this is, we had a disagreement about the proximity, what do we call it?

The proximity thesis, what do we call it? Hypothesis. I can't remember what we call it. We had a name for it.

And I think that I think that there could be a breakthrough of some kind, conceptually, of understanding what parallelism is. And I could be wrong. You know, maybe I'm just wishful thinking. You could be right that that, that bigness is by definition a problem because of the speed of light.

I don't know. And that this is again where it's so hard to speculate what, what the future really can hold with breakthroughs. With quantum computing, does this be light, is this be light still the same barrier? Well, did you see the news that the Chinese have quantumly entangled?

What do you entangle? Bits or what are they? Adams? What do you, what do you entangle?

What is that? I'm like demonstrating my ignorance here. But 750 miles apart, these particles were beamed down from space or something. Well, that, I mean, quantum could, could make huge changes because there seems to be, there seems to be this action at a distance that potentially against around us, potentially.

Yeah. If we, if we understood it. Sure. Well, I certainly don't.

So you know what I think that, what do you think? Well, my five year old, I'm a coma five year old because he, we were celebrating his birthday today with a birthday party. It's early where I am, came downstairs and grabbed himself at Dr. Seuss book and is snuggling with me on the chair where I'm recording the Skype, on Skype.

Yeah. And it's, it's just so yummy. It's just so delightful that I think I just need to live in the now and let these worrisome future scenarios play out the way they're going to play out. But we just enjoy the Goldilocks world that I live on where I can, you know, walk around outside and breathe the air and wear shorts in the teacher, no extreme weather and no predators.

And it's just really a golden age right now. Yeah. Any of our longtime listeners are going to recognize the yo-yo that we have been writing from the very beginning of the show between, you know, wanting to just enjoy today versus being terrified about, you know, the near future and what's going to happen for our children, our personal children that live in our houses, but then collective children. It's like, boy, I just had a realization that the simulation hypothesis has to be true.

Oh boy. Because if the simulation hypothesis were true, what would you simulate? When you want to simulate the golden age? Well, yeah, certainly golden for the people who have the resources.

Like old for everybody. Well, that, you know, I don't know that there's another big philosophical conversation. Of course, that's certainly true. And it's a lot more golden for everybody than if our atmosphere was a, you know, sulfuric acid atmosphere.

True. Or if we were so far, you know, we're definitely within the realm of human experience. There's, there's variability, but that realm of human experience is still in this week's spot of what's available in the universe in through time. All right.

Well, we'd love to hear from you, dear listener. You can email your audio feedback. That's the best way to advance this conversation. Send us a voice report on your recording phone.

We'll let you concern AI at gmail.com. And we will have a future show where we delve more into your thoughts. We are starting to get a backlog of listener feedback. Some of it, if you write it out, it's often very difficult for us to get on the show.

There's too much to read. It's too long. And then we have to use our people's human brains to kind of summarize it. And that's difficult for us.

So if you can summarize it in audio form, that's easier for us. So send it off to us. If you want to be involved in conversations in between episodes, you can go to Facebook and for the concerning AI group. Lots of smart people there, posting links to the conversation.

And until next time, keep your coffee up. Keeps good up. Hello.

Frequently asked questions

When was this Concerning AI | Existential Risk From Artificial Intelligence episode published?

This episode was published on August 8, 2017.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this Concerning AI | Existential Risk From Artificial Intelligence episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.