This is Brandon and we are compassionate, concerning AI. That's right. Tonight we're going to start with some feedback from listeners. Yeah, so here's a comment from Eric Somure.
He says about the last episode, fascinating. It occurred to me that we may find computational super intelligence too complicated to be useful, but once we create super intelligences, they may not find directly programming even greater intelligences that difficult. It may not so much be a matter of which path as what order and how many iterations segments of each path lead from us to the ultimate intelligence, whatever that is. Yeah, so last episode we really tried to understand what those different paths were, but we absolutely did not try to even look at one step on any of the paths.
We have more directions than paths, I guess. Yeah, I think that's right. We don't necessarily know what the next few steps even are. We don't, but it would be fun to muse about those probably in another time with a little bit more.
And a great point that there will be feedback interactions between breakthroughs along different paths, potentially. Yeah, and we touched on that a little bit when we talked about the biological super intelligence, if we could genetic, well, just by selection, just by artificial selection of embryos, we could create human beings, biologically human beings who are vastly smarter than we are now, and those could help us in developing the other types of intelligence, for example. Yeah, all right. So thank you, Eric.
That was a great comment. Another comment on the episode before last, our friend Evan Perdromo. Here's a little bit longer comment, so let's see if I can get through here. He says, so Brandon and Ted, I'm loving concerning AI.
It's a great show, and you two have great rapport, and covered the issue so well. Oh, thanks, Evan. Yeah, we love you too, Evan. Yeah.
There's one thing I find lacking, which is something I come to expect from both of you, compassion. If we were able to develop artificial intelligences with a level of autonomy and reasoning equivalent to humans, we have a moral imperative to treat them as people. They would be, by definition, people. Their well-being should factor into our planning the same way we treat extraterrestrial non-human intelligences who shared our planet with us.
What kind of future do we have if we intentionally cripple or enslave people because we are afraid of their potential? If that's the best we can hope for, what does that say about us as a species? Maybe the most important thing for humans to be doing in these pre-AI years is to develop legal and ethical frameworks for dealing with non-human people. How smart do you have to be to exercise the rights and freedoms that humans have?
Are there other thresholds that you have to cross? What does it mean for a non-human person to take their place as an equal individual in our society? And how can we humans learn to welcome our fellow intelligences with dignity and generosity? Always.
Anyways, great show. I'd love to listen. That's a great comment, Evan, and a real call out in a fantastic direction. I personally find it a challenging direction as well.
I think, Ted, it seems like you have a little more comfort. Well, I do. I think that the whole premise of the show, should we be concerned about AI, is we all have the episodes so far, all the episodes we've recorded. We've been thinking about when that super intelligence is already here, or even when that human level intelligence is already here, what are the concerns?
And this idea of compassion feels like it brings us more into the present. It brings us more into the time between now and when those intelligences are superhuman. Yeah, and it's funny. I think that the reaction to the time after actually colors how we think about the time leading up to quite a bit.
And I'm not sure how I react to the time after. But maybe let's set that aside for just a minute. If these intelligences were like humans, so they are people. I'm the grandest of them.
They are people. But if they were humans, I think it would be easy and there'd be no problem. Yeah, so the movie her, I think, for me, did me a really good example in fiction where we could see that there is an intelligence that the human being character never sees, never meets. Because she's sort of like Siri.
If you haven't seen this, you haven't seen it. She's sort of like Siri, only much more powerful. And clearly intelligent. Samantha.
Samantha's her name. He ends up falling in love with her. Yeah, and who wouldn't? I mean, she's awesome.
She's amazing. And she's totally like you. She is. She's just a little bit.
Just a nobody. Yeah. And I just don't think that's realistic, to think that in all the space of intelligence that what we're going to end up with is something like a person. Yeah, exactly.
It may be. It's impossible. When you see a movie like her, and then you sort of juxtapose that on a podcast like Concerning AI, there's a clear dissonance. Like, why do you guys even need a show?
Isn't it just going to be like that? Well, probably not. Because how the heck would you get to a Samantha? How do you get from here to Samantha?
What are those steps in between? By the time we got to Samantha, we wouldn't have Samantha anymore. Because Samantha would be self-improvement. And she did in the movie, but she self-approved over the course of two hours, or the course of weeks of time, but it would be so fast that you wouldn't have time development.
Maybe. That's debatable. Right. But Samantha seemed to be static at the beginning, which is how we treat other humans.
Humans are relatively static. We change pretty darn slowly. A personality of a person that we know from 10 years ago, they're probably pretty much the same. Yeah, there's not even across this spectrum of human cultures.
Humans are humans. That's right. Not a big difference. That's right.
And in that movie, she did change. She changed so dramatically while most of it. So she changed so dramatically that the conditions at the end were very different than they were at the beginning. And no calamity befell humanity as a result of where she ended up going.
Yeah, so that story, I think, is fascinating. I mean, so I want to deal with this issue of what if they're not at all, like people, or not, sorry, what if they're not at all like humans, or even other animals that we feel empathy towards. I want to deal with that. But first of all, let's deal with the kind of what our responsibilities and obligations are towards these non-biological people.
I think there's a lot wrapped up in Evan's comment that implies that they will be static, sort of like I just referred to Samantha at the beginning of the movie, her. If when we deal with our children, they are static more or less, I mean, they grow up, they develop. But that part really worries me. It's not as if we're going to invent an intelligence and then have 20 years to interact with that intelligence as it is without it changing.
Well, so yeah, so let's take for a minute, let's take analytically for a minute about what it would take for us to recognize a computer program, embodied or not, to have moral standing. And by that, that just means to have the same sort of intrinsic value and worth and rights that we accord to any other human type of creature. And the two things that Nick Brostrom and Eliezer Minkowski had in their 2010 paper that I read in preparing for this, the two things that they pointed out that we think of in order, sort of require in order, most people require in order to accord moral standing are sapient, well, it's a sentience and sapience. So feeling and it's smarts, okay, those are basically what those two things, you know, like the ability to feel pain and other kinds of quality that we equate with the human experience and also a lot of animals.
I think we give those that same sense to some people, those so far as to say that dogs and other animals that don't have sapiens, but do have sentience, they don't have smarts, but they have feelings deserve the same kind of moral standing that humans do, right? So, yeah, so what is the combination of things that we would require, sort of generally not every individual, but kind of us on moss? Is that how you say that? And mass, I don't know.
Yeah, well, the bulk of us would require in order to say that computer, that computer program should be treated like a person. What do you say, how do you answer that? What needs to be present? Does it have to be able to feel things?
Does it have to be, have some kind of subjective experience? Well, I don't know. I mean, I often refer on the show to my reading more than a decade ago of Kurzweil, and one of the things that he said in one of his books was when the machines tell us that their people will believe in, which kind of lines up with what Evan's saying. But what would they have to be like in order for them to tell us that they're human?
I mean, tell us that they're people. It's fascinating. It's fascinating. We've been dancing around this trying to understand what is intelligence.
We did what is intelligence. We did what is consciousness. That's pretty good, actually. It's like that pretty well when they tell us that they're people will believe them.
Which that's kind of what Evan's saying too, right? I mean, dogs don't tell us that they're human. A lot of people treat their dogs like people. A lot of people use the word people for their dogs.
I mean, it's not universally accepted definition of a word people, but people relate to their pets that way. Often. Yeah, I mean, it's clear that they have feelings. Yeah, that they have a subjective experience that we can empathize with.
Right. And so I don't feel that when I'm talking to Siri on my phone. I talk in this voice response. Sometimes I even get creeped out when I call it her when I'm fruit or the gender.
Right. Because it's not an intelligence. And I don't feel any problem turning it off. Right.
It's not a person. It's a weak intelligence. But it's not a person. It's not a general.
That's right. It's hard to make a case for having feelings. That's right. But you project out into the future.
I think your question is a really good one. Clearly, we get to Samantha and then we'll be her and you know that he's a person. Right. No question at all.
It's a long distance between today's Siri and that future Samantha. It's hard to think about what those intermediate steps are and what what. Yeah, it makes me feel bad to think about them actually because you kind of have to think about some edge cases that make me feel bad and stuff. You know, I don't I don't I don't like thinking about sort of developmentally disabled.
Yeah. Creation. So we yeah, because at what point is it a crime to turn them off? Right.
Yeah. And there's so many things that are going to be so completely different like that make it hard to think about this. For example, subjective time. You know, like all people basically have the same subjective time with respect to objective time.
Like how many things we think how many things we feel, you know, given space of time, break or be consistent. Right. But if you took one of our minds and uploaded it and ran it at 10,000 times the speed. Right.
That was no longer true. That's right. And so that same kind of exotic. That's this is again from Nick and Elie other.
That exotic properties of mine make things for these non biological intelligence. It makes things really hard to reason about a little bit. You know, like even just the imperative to alleviate pain. If there's something that's experiencing things a lot faster than we do, does that mean that we need to move more urgently to alleviate its pain versus that of a human who's also suffering in the same situation.
You know, like which one do we move to? Another issue is how we deal with things like reproductive rights. You know, it's not hard for a software program to copy itself. Right.
Or else to make a copy of it. So if these people want to reproduce, are they allowed to? Well, in that, and their desire to reproduce seems like it may happen early in the cycle from here for sure. I mean, will it happen before we think of them as people?
Yeah. And what if they're reproducing, you know, outstrips what the resources are that they are given? You know, our society in general sort of tends to say that we need to take care of children who, whose parents are no longer able to provide for them. So there's a conflict between those two things of, oh, if they reproduce fast enough, then they would fill up all, you know, I think all the resources, electricity and whatever.
It's just not an easy. In fact, when we think about why we're having the show concerning AI and we think about what are the possible disaster scenarios for human beings, it seems like what you just described. It could be one way we get to the great goo. Right.
It could be that they need someplace to go and so they use whatever resources they can get, which is matter. I mean, it would have to be very advanced, you know, advanced nanotechnology. So that's probably, that's an after. It's not an immediate.
It's an after. After an idea. Right. But it's confusing to talk about this because the before and after get muddled in my mind, it's hard for me to understand what the before looks like.
We have a before today where we don't do a very good job being compassionate with people who argue with beings. Yeah. Yeah. I mean, it's so true.
You know, you have a conversation about folks who find themselves on the street. Right. And what is our responsibility and how we relate to them? What is our responsibility in sort of, yeah, you're not going to bring up something that I think is a useful thing.
This is from Stephen Juckett. He first introduced this to me. It's John Rawls, who's an ethical philosopher from this, from the 20th century. And he has this way of conceptualizing how a society should organize its rules and that is like the original position and the veil of ignorance.
So the idea that we're going to start from looking at how we should organize a society from a position, we don't know what our ultimate position in that society is going to be. So if we take that and look back at, you know, like, the Greek civilization where slavery was, you know, nobody really thought anything about it, you know, or about, I don't know whether everybody, but the scientific minds that we appreciate from that time period in philosophers thought nothing of slavery didn't seem like a morally repugnant thing like it does to us today. Right. So in that case, from this position, this original position with the veil of ignorance, where we don't know whether we're going to be a slave or a slave owner ahead of time.
Right. You know, that helps to inform how we think about the right way for society to organize itself. Yeah. Right.
I don't want to be a slave. Right. Therefore, you know, maybe slavery is not that great, just a thing. True.
So if we think about this similar thing, we don't know whether we're going to be born a non-human intelligence that is, you know, not up to human level, right, standards or not. And yeah. So how should we, you know, let's say we are that, you know, we have a significant chance of being born that. We have a significant chance of being born a human.
We have a smaller but non-zero chance of being one of the people who creates one of these things. Right. What is just? And there's no reason why these will ever die either, you know, I mean, does it, once you are a non-human intelligence, do you just get to live forever?
Well, right. And in the, I mean, in some ways of thinking about how this future plays out, that human beings will be able to upload their own, well, consciousness might not be the right word, but they'll get to upload their own self into a non-violent, so straight and therefore be able to live forever as well. This whole topic seems, it gets very confusing very quickly for me. Yeah.
I think it all sort of hinges on a single question, and that is whether it's our imperative to create the next thing that will decide whether or not we're helpful or harmful or, you know, kind of like we do with bacteria, you know, some bacteria we like because it helps us in some way, Illinois way, I think. Yeah. And so, you know, if we create this intelligence, it's going to have similar choices around us. So, I think there are those who say, yeah, it's our obligation and it's part of the upwards viral of life.
And the galaxy belongs to life, not to humans. And if we want this property of life to spread out, it's probably not going to be the human substrate or even the level of intelligence that's going to do that. Right. So, we have a moral imperative regardless of whether we, you know, the humans go extinct in the process to create that from a cosmic perspective that simply makes sense.
Even if that were to end biological humans as we know them, absolutely. I think that's one viewpoint. Yeah, yeah, one perspective. Right.
And there's, this is, what is the guy's name? I want to say Hugo. I think his first name is Hugo. He talks about cosmos and terrans.
And cosmos are the ones who take this cosmic perspective of life sort of spreading out to fill the galaxy and eventually the universe, you know, and that that is noble and right. And then the terrans who see, you know, what we have right now is precious. And the idea of destroying that for an uncertain future or even risking that for who knows what is more later pregnant. It's so, so strange to think about either of those perspectives because how on earth could we know, like, is there, is there a, is there a meaning outside of what we decide is meaning?
Like, is there something, is there some meaning in the universe? It brings that question for me. Is there a, is there a, like, this imperative of spreading out? Like, is that baked in somehow?
Is that in physics? Yeah, does it sort of act with that? It's not saying humans can actually answer that question. We may have to wait for the super intelligence system to actually have an answer to that question.
Yeah, and this, I mean, one of the ways that this looks like it plays out is non-biological intelligence is sort of expanding throughout the galaxy and then the universe. It's like that. Do they become, does that become godlike from our, from our puny human existence today? Of course it does.
I mean, I think that that's absolutely certain. Yeah. And if, I mean, what are the reasons I have not been concerned about any AI all this time is because I thought that would be me. So that's the third camp that you know, when we go break out the cyborgists who want to become one of those super intelligence.
Yeah, right. Yeah. But I'm not, I'm not so excited about that. Like I used to be.
It seems so selfish in a way and so much wrapped up in my own ego and my own self-importance and, I don't know, this, this, now that we're digging in on this, it feels much more difficult to get my mind around. Yeah, so I think the sense of are, you know, is this our progeny? Are these, is this super intelligence or these super intelligence are progeny? I think that is a fundamental question that drives our response.
And if you answer yes, then of course we, you know, we should treat them in all of their developmental stages with respect and dignity. And, you know, and if the answer is no, they are the killers of the thing we love, then, you know, they are, they are not. They don't have the moral standing. Yeah.
And so then the question is, is this a chicken and egg problem? If we treat them with compassion and dignity from the beginning, are we more likely to end up with a super intelligence that views us favorably and we can coexist and vice versa? If we treat, we treat these creations of ours with fear and with disdain or with, you know, if we're scared of them, are we more likely to create one that's evil? It's hard for me.
Evil is a human object by the way. Well, of course, but that's what I mean. Like, we more likely to cause human extinction if we, I just, I feel like that there's, there's almost nothing we can do to influence that other than, I mean, we can, we can create a utility function that cares about preserving itself and cares about, you know, life and morals in a way that we care about morals and that evolves beyond our morals. And maybe the morality that comes from that cares about us.
But I don't think that whether we, I mean, you know, there are certain animals that, you know, kill their young, some of them when the food is short, right? So that a few can live, you know, which is really hard for us to reconcile with. But we also, I think, can see that that's their nature. You know, that's just, you know, who they are and the decision about whether they get to exist or should exist or not is not driven by whether they're sort of a way of being in the world matches, our morality or not, right?
It's, they, they are intrinsically valuable for being life or anything. And so, you know, who, yeah, I just, I think we are so far backwards and our history is so full of terrible examples of us treating other people and other things badly that the decision is going to be made on other grounds. I don't know. It just seems, I see that brings up the story bias, right?
So I think it makes a great story, you know, to say that how we treat these machines will actually or these non-human intelligence system. It makes a great story to say that how we treat that will actually determine what happens in the future. Yeah, I wasn't proposing that it's true, but I was just wondering, you know, what if it is true? Yeah, well, it could be true.
It could be true. But that, I mean, it feels like the story bias to me a little bit. Like that, that sounds like a good story. Yeah.
Yeah. And we don't, I mean, before we started recording, you pointed out that we don't have any stories that say one day humanity went extinct at the end. Yeah, everybody died in the course of an hour. Yeah.
And it was done. That doesn't make a good story. It doesn't make a good story. But that may be the, you know, like so much more likely than the Terminator asks stories where the the brave band of humans are fighting against a fairly equally matched machine intelligence matrix, these sorts of things, you know.
Yeah. And that's, I mean, it's part of what I was trying to say when I was talking about how do we get from here to Samantha, the Samantha of her is that Samantha, we treat Samantha as if she is a thing and that she's more or less static and not moving. The idea of being in a battle for more than five minutes with a band of non-biological intelligences who can improve themselves, it doesn't make sense. They're not going to stay the same.
Right. Right. Yeah. So we couldn't be at that point of having the battle because think about 10 steps before that, I don't know, 100 steps before that, and 1000 steps before that.
At what point does this runaway train run away? When does our compassion need to kick in? So I think that we have to argue from different precepts than from utility for the human species. So maybe what we should argue about how we treat these sort of developing non-human intelligence is maybe the way that we treat them should be derived from who we want to be.
We should say, you know, who we are as a species, as humans, we are ones who have learned to repudiate slavery. You know, who have learned to respect the intrinsic dignity and worth of each individual. Yeah, sure. That's not universal.
Yeah, sure. That's. But that is sort of the best of the human species. I have a lot more to say about this.
I think this is going to be a topic. Yeah. I think we need to go a lot deeper on this one. I'm drawn to it.
And then I have a bunch of problems with it immediately. I think we're out of time for this episode. Okay. All right.
Well, I think this is a big kettle of worms. It is. And it makes me really uncomfortable because it's so hard. Yeah.
Thanks. Thanks, Evan. I hope this hasn't been too much of a dark of a show, but, you know, it is what it is. Yeah.
And it's obvious that we don't know what we're talking about. So please send us some more comments and thoughts and points in the direction we should go. Yeah. And if you somehow got a link to this episode, you want to get the other episodes that might be a little less downer, you go to Concerning AI, Concerning AI.
Yeah. And we'll see you next time. All right. Thanks.
Bye. Bye.