This is Ted, this is Brandon, and we are talking concerning AI here. Today's topic is skeptics, and we're going to talk about what that means. What does it mean? Yeah, there's a couple of varieties of AI skeptics.
There are those who think that the time frame is wildly optimistic. So I would say, let's classify those as it won't happen in this century. The people that think it won't happen in this century, they are not concerned about AI because it's not happening now. It's not happening now.
All right. All right. What's the other kind of skeptic? The other kind of skeptic is the person who says, yeah, it's going to happen this century, but we don't have to worry about it yet or.
So they're still with that first group on the timing of it. But then there's another group of people that is completely different. They are a skeptic in that they say, it's not going to be a good thing. They're the enthusiast who say, AI explosion is going to happen, and it's going to be Nirvana, it's going to be great.
There's no way to think of it happening any other possible waycarts file. Then there are those who say, it's inevitable and it's going to happen and I'm really worried about it. There's no reason to think it's going to be a good thing. So I'm skeptical that this positive future for AI is going to happen.
Yeah. So from our perspective of concerning AI, it seems like we need to assess in some way why do some people think it's going to happen right away and other people think it's going to happen a long time for now, but both think it's inevitable. Then what is the likelihood that it is going to be Nirvana or that it's not? Yeah.
I wish we could split the term skeptic and really distinguish between these two cases. So maybe we should just call one, not now. Not now, skeptics. Not now, skeptics are not soon skeptics.
Yeah. Because I think we could really drill in on what is the distinction there. Yeah. And then the other kind of skeptic, it's almost the opposite.
The people, let's not use the word skeptic for them. Let's use the like warriors, like enthusiasts and warriors. Right. Because if we are concerned about AI or we're trying to assess, should we be concerned, we want to know who else is concerned.
Right. Who's worried? Yeah. The warriors.
I think that's a good thing. Should we talk about timing first? Yeah. Let's talk about the, so let's call those people the skeptics.
For the purposes of our podcast, the timing people are the skeptics. Okay. Is that seeing fair? I think it's fair.
That's why I thought it before we started reading this and encountered all the people who are putting up the enthusiast versus the warriors as the enthusiast versus skeptics. Right. So let's just use it the way that it seems national to us. Okay.
So when we say skeptics, we're talking about people that are saying that AI explosion is not happening right now. It's not happening. And that lets us then have a spectrum like, you know, Jaron Laneyare, which I'm not sure I'm pronouncing his name right, but he's at the sort of low end of that group. He's like, hey, it's, you know, it's not going to happen anytime soon.
Don't worry about it. But then when pressed, he's like, of course, it's going to happen this century. So even though he is still kind of in the area that we think is concerning this century, he's still in the same posture. He says the others.
We don't need to worry about this now because it's not happening for a while. Yeah. And he actually says in some far future situation, it's far future situation we're going to transition. But then he also says the century.
So there's a disconnect in my mind between those two statements. Right. When he says some far future, that makes it seem forever in the future, right, you know, indeterminate. Right.
It doesn't, like the lives of my children, their lifetimes is not seem far future to me. Well, that's right. And this is, this is definitely within the lifetimes of kids who were born at the same time. Even, even us in our positive waters, without radical life extension, there is some chance that we could live to see the next, the turn of the next century, you know, even without a huge medical breakthrough.
You know, it's like, yeah, not most people don't live into their 120s, but it's possible. Right. Right. With just some incremental extensions, like, like that we are seeing happening right now.
Certainly the lives of our kids, which is why we're sitting here making the show. Really, that's how we got to think about it was with our kids. So, so that feels weird to me that he, that he talks about it like that. So that, that was a fascinating thing to me, was how really the people who say it's going to be outside of the century are very, very few.
Even among the most vocal skeptics, the ones that are held up as, you know, we don't need to worry about this. Most of them are still placing it this century. Yeah, which, that's greatly concerning to me. That's great.
That's what I mean. The one person that really, um, I'm going to have to dig into more because I, I really respect the field that he's kind of the figure head for, is, is Rodney Brooks and embodied AI. You know, he's putting it at maybe thousands of years. It seems like that may be a show for us.
That may be a topic for a show is to really get into Rodney Brooks, uh, timing, embodiment, a whole bunch of things that he's talking about. I don't feel like we're really equipped to dig into that now, but that, that feels like a future topic for us. Yeah, I, I, it's fascinating to me, you know, when I think about the different, the different thrusts of AI research, the one that seems to me like, uh, is the only path, really, that I can see general intelligence coming from, is the embodied path. Because so much of the way that thinking, uh, the way of thinking about the world, which is what we're worried about, something that actually interacts with the world we live in.
Yes. So much of that's grounded in perception and, and, um, sort of perception and affecting or manipulating the world. So much of our language, so much of our metaphor, so much of the way we think is absolutely tied to our perception and our embodiment. Yeah.
I think that's going to be another show, more important about it. We did, we touched on a embodiment back in one of our negative episodes, negative three, maybe, we touched on it. Yeah. But I would love to go further.
So maybe, maybe in a soon show. So I guess the point I'm trying to make is somebody who's, who I, judgment, I respect, Rodney Brooks, uh, is not worried. That, I, the problem with that is I hope to find a lot more people. There weren't enough non-concern people, even the ones who say they're non-concern because it's so far out there, their time horizon still generally is pretty short in our view.
Yeah. Yeah. All right. Have we covered time?
I think we've covered time. I think we've covered time. Uh, I guess we should probably try to quantify things a little bit because, you know, again, Jaren, I wish I knew how to say his name, uh, linear is his last name. This is last name, linear says, you know, hey, it's in some far future situation.
So unquantified. Right. And then when you push him for a quantification, uh, his take on far future is very different than what we would say far future. So let's quantify two things.
Let's quantify our sense from reading this of how many people are in the, won't happen this century of, of these people who work in the field or are smart technologists. How many of those people are saying far future won't happen this century and how many are saying it's, you know, got a 90% chance in my estimation of happening this century. And my sense was that that was like maybe 10% are saying, but it's outside of this century. I think that that's right.
I think it's a, it felt like a vanishingly small number. Yeah. And that may be my own perception minimizing that. So 10% feels, feels like a way to combat my own bias that it's smaller than it is.
I think, I think that might be generous. I think 10% might be generous. Yeah. I think it might be too.
Uh, but that's the sort of thing we're talking about. So, so even on this continuum of long, you know, now, those who would, who would self identify as we don't need to worry because it's happening in the far future, that's a much, much higher percentage. Yes. But therefore future is not our far future.
That's right. I think that, I think that is very accurate. So maybe 50 50 on that. That's right.
People are identifying as no worry now because it's in our future. I think there's another group that we might want to try to quantify. And that is how many feel like we could predict this with any level of certainty. It feels to me like, and I'm thinking about a quote from one of the articles, John McCarthy once told me that when people ask him when human level AI will be achieved, he says between five and 500 years from now that statement feels like an outlier from everything that I was reading that people are not feeling as uncertain as that they're feeling more certain.
It's interesting. I kind of agree with that statement. Yeah. Yeah.
And so, you know, just because it could happen anywhere between five and 500 years from now doesn't mean if I were to draw a probability distribution of when I think it's going to happen, right? That that wouldn't be peaky, a peaky distribution, you know, in the like, say 30 to 70 year timeframe, right? Right. Even though it's non negligible.
You know, it's got a long tail that is substantial. Long tail both directions. Yeah. No, it's not going to be a symmetrical around the middle.
It's the long tail is more on the far future. Right. So I think that that statement is perfectly consistent with John McCarthy believing, I don't actually know what he believes, but him believing that there's a 90% chance we'll have in the century. Yeah.
I think that's probably right. What it also brings up for me, I wish I had some current math chops. You know, I used to be, I used to be able to do math a long time ago. I haven't done math in a long time, but I would love to be able to look at what are the parameters that need to change very little in an exponential curve in order to generate a distribution that's five to 500 years.
You know, thinking about something that could be 70 years, but that with a tiny change today, we turn into 30 years or turning to 150 years. Like, I'm thinking about, I'm thinking about the math and how the exponential overcomes anything. Yeah. Well, that assumes that, you know, that the Kirk file, who's so confident, is in a camp that I think is unsophisticated.
And that camp says that all we need is computational resources, but the programming is not actually that complicated. Right. And I don't think that that that our understanding of the human brain bears that up. And so I actually think that we don't know how to use this.
We don't know how to use the computational resources if we have them right now. And so it's going to take some breakthroughs of some sort that we don't know and can't predict, which is why it could happen a long time from now. I guess what I'm pointing to is if the slope of the exponential is much flatter than exponential. Well, the exponential of all of human progress.
I mean, Moore's law is the easiest way to think about it. And that's silicon. It's very specific. It's very narrow.
But I think there's a strong argument that says that Moore's law is only one specific of a generalizable trend. Another example that Kurtz-Baupt points out is our ability to the specificity of which we can measure and view the human brain. And what's happening here, it's the other direction, like we're getting smaller and smaller resolution. But if you generalize Moore's law to the number of floating point operations or some sort of operations per second, that's continuing, even though we can't minimize your interest or as much more.
So those kind of simple exponential curves are completely different than fundamental breakthroughs in how we see the world. If Einstein hadn't come up with relativity when he did, it could have been another 20 years before somebody came up with it. It's about having a flash of insight that is inaccessible to most of the world. So you have to have somebody in the right place at the right time thinking the right way in order to make those connections.
And it's possible that in AI that there's going to be a few fundamental breakthroughs like that in how we think about the world. And if those people are in the right place thinking about it at the right time, they happen sooner. If not, they happen later. It's not the simple exponential.
I agree. And I think in retrospect, when you look back, those things seem to be smooth out somehow. Yes, it's true that 20 years, and that's kind of what I mean. If there's a breakthrough that we think we need in 2015, but that doesn't come to 2035, by the time we get to 2500, that doesn't matter.
So yes, our lifetimes, a 20 year difference is huge. And that's kind of what I'm, does the slope of the curve change or is it just where the curve starts? I see what you're saying. And this is where I'm struggling a little bit with not feeling like I have the math jobs right now to kind of understand what do exponentials really do.
And we stack S curves. Well, we're running out of words a lot of transistors wise. We're getting close to the laws of physics, I think. So that's next.
Yeah, we're getting close to signaling pathways that are on the order of an atom in their width. It's problematic when you try to think about the future of that. So, so you know what's interesting to me in this is why is there the distinction between those who say, you know, don't worry, everybody. And those who say worry, worry, worry, worry right now.
Because to me, their positions about when this is going to happen are virtually indistinguishable. Well, that brings up a quote that we both found insightful. Should we read it? Yeah.
The skeptic position seems to be that although we should probably get a couple of bright people to start working on preliminary aspects of the problem, we shouldn't panic or try or start trying to ban AI research. The believers, meanwhile, insist that although we shouldn't panic or start trying to ban AI research, we should probably get a couple of bright people to start working on preliminary aspects of the problem. Yeah, I mean, this is saying it's the same. So there's a different emphasis here, right?
There is. One, I think, is rallying the truth, saying, hey, everybody, let's get busy on this a little bit more than the other one who is saying, don't panic. So do you think this has to do with the hype and the worry about funding and, you know, there are those who believe that we're setting AI up for another bust? A hype has been a big problem in the last 50 or 75 years of technology.
We're still waiting for our flying cars, right? Or fusion, which is always 30 years away. Right, right. And so, yeah, we've had all kinds of hype and AI.
What the heck is AI anyway? I think let's touch faces with Jan LeCoon here. He's the head of Facebook's AI lab and is generally held up as one of the skeptics. Yeah, what does he say about hype?
Well, and I think he says, hype is dangerous to AI. Hype killed AI four times in the last five decades. AI hype must be stopped. So I think that is what is motivating most of the people who are in the far future camp.
We're saying, hey, it's going to be a long time. Don't worry. It's going to be a long time. Don't start getting excited yet.
And definitely don't start worrying. Yeah, don't start worrying about it yet. And don't start imagining that this is going to happen anytime soon because, you know, on the order of a human attention span of you funding agencies and you, the media, it isn't going to happen anytime soon on your order of attention span. And I think they're probably right about that.
On the order of attention span of the media and of funding agencies, it isn't going to happen anytime soon. You know, 10 years, 20 years is a long ways out in those attention spans. It is. It is.
But I guess that brings up for me the concern, which I don't know if it's a valid one or not. But if we were to wait 10 or 20 years to really start looking at safety, then the train might have left the station already. Right, it's possible. It's possible that we're setting up initial conditions that will be ultimately be disastrous.
So I think I definitely, as we're talking about this, I definitely fall in the alarmist camp for two reasons. So one, if it turns out that, you know, it is going to happen sooner than we think, it would be good to have given it our best go around early. Yes. That's one.
Two, we can actually potentially push it out, which I kind of in the camp of, it would be good to push it out because it's so uncertain what happened after it. That when you say push it out, you mean go slower? Yes. Push out the explosion, the intelligence explosion.
Right. As far as we can to give us a little more time. So maybe having a hype and bust is in our best interest in species and for, you know, DNA based life. Yeah.
We might need to have a show based on who's doing AI research and where's the funding coming from. We might need to think about that. You know, as I don't know whether a hype and bust is going to slow down in Google, for example. That's true.
The other billionaires that are out there are funding this stuff. It's like, they almost don't care. Good, good point. So maybe that would be interesting to see who is, if funding source is correlated with these two camps, that would be really interesting.
That would be fine. I wonder if somebody's done that. If you know if somebody who's done that, we would love to find out that correlation. How do people get in touch with us?
I don't know that we've told anybody how people get in touch on the show. Do we have any email address for the show? We do, but it's not working. So let's not advertise with this show.
Let's put it out on the show. So you can get in touch with us in some other way. And next show will tell you for a while. Yeah, maybe Facebook is the way to go for now.
We have a Facebook concerning AI group. That seems to be the most active place people are so far. Sounds good. So if you want to get in touch with us, let us know there.
If you like the show, go to iTunes and give us some stars. I don't know if anybody's done that yet. So you can be the first. Wow.
Wouldn't that be cool? I love stars. All right. Well, we don't know what our next topic is.
So look out. It's going to be exciting though. So subscribe to the feeds we want to look for. Yeah, then you won't have to look for it.
So show up when we're ready. Super. I'm Ted. No, I'm Brandon.
No. Anyway, we are concerning AI. See you next time.