ANNIE MINOFF: This is Science Friday. I’m Annie Minoff. You may have heard of the term the singularity, the idea that in the next stage of evolution, humans and machines will merge. The concept was popularized by Ray Kurzweil in his 2005 book The Singularity Is Near, which has been essential reading for technology buffs and critics since its publication.
Well, that was 20 years ago. In that time, we’ve seen huge advances in artificial intelligence, computing power, and research. So it seems like we’re due for the next text to bring us up to date. Lucky for us, we’ve got it. Here’s Ira Flatow with more.
IRA FLATOW: Joining me now is futurist and inventor Ray Kurzweil, author of the new book The Singularity Is Nearer– When We Merge with AI. Ray has worked with artificial intelligence in that field for more than 60 years. I’m going to guess longer than probably any other person alive. Welcome back to Science Friday, Ray.
RAY KURZWEIL: Yeah, great to be here again.
IRA FLATOW: Nice to have you. When you look at the last 20 years, is it hard for you not to just, yeah, I told you so? I mean, has it gone the way you thought it would?
RAY KURZWEIL: Well, other people are saying that. In 1999, I predicted by 2029, we’d reach the Turing test and also AGI, where basically a computer can do anything a human can do. People met that with a great deal of skepticism. Stanford had a conference. Apparently about a thousand people came from around the world. And everybody agreed that this would happen, but they didn’t think it would happen in 30 years. The average was a hundred years, and some people said several hundred years. Nobody said it would happen within 30 years, but now everybody is saying that.
IRA FLATOW: Well, why did you get it right and they didn’t?
RAY KURZWEIL: Because I have a graph, price-performance of computation, and it’s a straight line on a logarithmic graph. So as you go up the graph, it represents an exponentiation of the speed. So in 1939, we did 0.000007 calculations per second per constant dollar. And the last thing we did, 135 billion calculations per second. So that’s a 75-quadrillionfold increase for the same amount of money.
And I basically used that chart basically for most of my predictions. I figured we needed about a trillion calculations per second to achieve what human beings could do, and I figured that would take place in 2029, and we’re right on that track. It’s an absolutely straight line on a logarithmic graph. It’s pretty extraordinary.
And we were actually changing different types of technology. In the beginning, it was relays, then discrete transistors, then integrated circuits. No matter what type of technology we’re improving, we made the same shift in exponential terms every year for 81 years. It’s pretty remarkable.
IRA FLATOW: Yeah. You predicted that AI will pass the Turing test in 2029. Does that prediction still stand, or might it be even earlier?
RAY KURZWEIL: Well, most people now are saying it’s going to happen earlier. Elon Musk says we’ll do it within two years. So I’m saying five years. I see no reason to amplify my prediction, but it probably will happen sooner than that.
IRA FLATOW: And in terms of the singularity, you predicted it will happen in 2045. Is that still the case, or could that be earlier?
RAY KURZWEIL: I’m sticking with that prediction. That basically means we’ll be able to shift our awareness, what happens in our own brain as well as what happens in the cloud. And the reason that we don’t get smarter is we have a limited size for our brain, but if we can actually go outside of our brain and into the cloud, the cloud right now is expanding exponentially. It gets twice as fast every year for the same money. We’ll be able to think much deeper thoughts. It’ll feel the same except we’ll be able to do what we do now– we do it with our phones. Phones are a little slow. We’d like to actually just have these thoughts appear in our brain without having to wait to type things in and so on.
IRA FLATOW: Are you thinking of a brain-computer interface here?
RAY KURZWEIL: Yes, exactly. But if you go back to 2005, we didn’t even have phones. Very few people had phones, and if you had a phone, it did very little. I’ve actually done polls where I’ve asked people, who here has their phone? Everybody raises their hand. Nobody goes outside without their phone.
IRA FLATOW: And so we’ll have a brain-computer cloud interface that no one will think twice about is what you’re saying?
RAY KURZWEIL: Right. And people say, oh, I wouldn’t do that. But people didn’t think they would use phones 20 years ago either, carry around this thing and lose it, but now everybody has a phone. And we’ll be able to interface our brain wirelessly to the cloud, which is growing exponentially.
But a lot of things are happening now that we only just thought about in 2005. For example, we’re doing now biology this way. For example, Moderna, they actually gathered several billion different possibilities for what could cause COVID. And generally we would go through that with personal trials, but they actually did it with simulated trials, and they did that in two days using AI and different types of techniques to eliminate various possibilities and came out with one. And that’s still the solution that we’re using now for COVID. Was done in two days a few years ago.
And, actually, a variation of that is now dealing with cancer. We have our first trial with a thousand people in England where we’re testing pancreatic and breast cancer and several other different types of cancer using something very similar and getting very positive results so far. We don’t have the final results yet. But that never could take place before. This is actually the very first time that this has happened.
So we’re going to go through all kinds of diseases using simulated biology. As of two or three years ago, we had gone through 170,000 proteins. That was actually done by person, but that was a very small fraction of them. And then AlphaFold 2 in 2022 went through 200 million proteins and got the shape exactly correct, and that’s going to help come up with new medications. So there’s a lot of things that we had imagined or that I imagined in 2005 that are now actually coming true.
IRA FLATOW: You had also imagined, if I remember way back in the day, that people might live forever.
RAY KURZWEIL: That’s not exactly what I’m saying. We’re going to reach longevity escape velocity. Right now, you go through a year. You use up a year of your longevity. However, scientific progress is also progressing, and you get back a certain number of months from that. Right now, you’re getting back about four months a year. So you lose a year, but you get back four months. You’re only losing eight months. However, the scientific progress is progressing exponentially. Somewhere between 2029 and 2035, you’ll get back not four months but a full year. So you use up a year, but you get back a year, and you’ll basically stay in the same place as far as your longevity is concerned.
Now, that’s not a guarantee. You could have a 20-year-old, and you could compute their life expectancy as many decades, and they could die tomorrow, let’s say, from an accident, although we’re also dealing with accidents– for example, cars. Right now, we lose 40,000 lives from humans driving cars. When a computer drives a car, it’s going to be much safer. In fact, that’s one of the reasons it’s held up. We don’t want any accidents, and so we’re going to get rid of a lot of accidents in cars and in other places.
But it’s not a guarantee. You could develop a disease that we don’t have a cure for yet. But as we go through a year, you’re not going to be using up a year of your longevity.
IRA FLATOW: Because we’ll have new medicines and–
RAY KURZWEIL: Exactly.
IRA FLATOW: –vaccines and things and body parts.
RAY KURZWEIL: Yes.
IRA FLATOW: Because we wear out our body parts, right?
RAY KURZWEIL: Absolutely. In fact, I’m involved with a company where we’re creating artificial kidneys, and people are working on artificial hearts, artificial lungs. So we’re going to be recreating body parts as well, and we’re going to be making progress at a much faster pace than we’ve done before.
IRA FLATOW: As we watch artificial intelligence, AI skyrocketing, we’re also aware that it uses a lot of energy, right? It’s using more energy than Bitcoin mining these days. Can we achieve the singularity without completely depleting our energy sources and our grids?
RAY KURZWEIL: Well, we’re also making fantastic progress on converting renewable resources into energy. Right now, solar cells are 99.7% cheaper since 1975. We’ve increased the amount of energy coming from that 2 millionfold, and the total amount of energy we’re getting from solar and wind, it’s doubling every four years. So at that rate, in about 10 years, we’ll be getting all of our energy that way, and it’ll keep getting cheaper. So, ultimately, it’ll be very inexpensive.
We actually have 10,000 times more sunlight that falls on the Earth than we need to meet all of our energy needs. Some of it’s hard to get to, but we have plenty of headroom, and we’ll be able to more than do that within 10 years.
IRA FLATOW: Do we do we have enough headroom for making computer chips that all this stuff is going to need? I mean the basic elements needed for computing. What about mining of the materials we need for it?
RAY KURZWEIL: That’s all part of the process. I mean, it’s gone on for 81 years. We find the materials. We’re able to create many more computations per second. I include in the book an analysis of what the ultimate would be. We call that computronium. And we’re very, very far from that. So we can keep this exponential growth going for a very long time.
IRA FLATOW: There is a climate-change argument against the massive use of computing and AI. Can they work together?
RAY KURZWEIL: Well, that’s only true if we use nonrenewable sources, if we use oil and coal and so on that add to carbon dioxide. These new forms of solar and wind don’t use any of that. And we ultimately can go and use all of our energy that way and completely eliminate adding carbon dioxide to develop energy. That’s where we’re headed.
IRA FLATOW: Do you think we could ask AI to help us solve the climate-change problem?
RAY KURZWEIL: Well, it’s doing that by converting things that don’t add carbon dioxide to create energy, and that’s going at an exponential rate. So as we get to the 2030s, we’ll actually be able to reduce the amount of carbon dioxide we’re creating.
IRA FLATOW: Moving on to the dominance of the AI industry, do you think there needs to be checks and balances so that AI doesn’t get out of hand? I mean, how can we make progress without losing control in this race to the top, or do you expect that AI will take us over in that singularity?
RAY KURZWEIL: Well, yes and no. I mean, we do need some more guidelines. We actually had an Asilomar meeting because the Asilomar meetings on biotechnology have worked out very well, and we’re doing that now for AI. There is a lot of guidelines already. I mean, any of these major companies, if they put out something that causes a great deal of harm, there will be a lot of liability, so they’re very motivated to avoid that. And I would say in all of these new changes, they put more effort into avoiding problems than creating the new developments.
IRA FLATOW: So the insurance companies, as they always do.
RAY KURZWEIL: Yeah. Well, that’s actually part of the motivation. And sometimes people look at the problems we’re creating, and they can imagine the problems, and they don’t imagine any new solutions. And they go, of gee, if somebody put this out, we’d have a big problem. But we’re actually making advances in how to avoid problems at the same time that we’re creating these new problems.
So far, it’s worked out well. We’ve had this concern. I mean, the Luddite movement started 200 years ago. They were correct that things like the cotton jenny and so on did eliminate lots of these jobs. We created new jobs. And if you ask, well, What are the new jobs going to be? and you asked that 200 years ago, no one could answer it because we hadn’t even invented these new jobs and the new industries that they were employed in.
So I’m optimistic, but, I mean, there are definitely problems. Intelligence is really the most important resource we have, and if you have somebody that’s not of right mind who has a lot of intelligence, that could be a problem. So I’m not saying there’s no issues. I mean, we really have to pay attention to this.
IRA FLATOW: Speaking of paying attention, what about the upcoming elections, the possibility of phony AI, AI-generated statements and photos?
RAY KURZWEIL: I mean, that’s probably the biggest problem that AI is presenting for this election. It probably will take place right before the election, so there won’t be time to actually sort it out because if it happened now, we’d be able to figure out, OK, that’s probably a fake answer. I mean, you see Biden, and he says something. It looks real, but it’s completely fake, and that could cause a lot of problems.
It hasn’t really happened to the major players. So I’m not saying there’s no problems. I mean, we really do have to pay attention to this. I think we’ll make it through, but these are definitely issues we didn’t have in the last election, for example.
IRA FLATOW: Yeah, and as we move toward the singularity, what concerns you most about that?
RAY KURZWEIL: Well, we’re going to have to make changes much more quickly. You ask, actually, people– and I cite polls where people think that poverty is getting worse. 80% of the people think that poverty is getting worse, whereas actually it’s been reduced 50% in the last 20 years. And so we’re actually doing much better than we thought, really because of technology.
But we had time to figure that out. I mean, how many jobs were lost from the railroad? But it took decades for that to happen. Now it’s happening very, very quickly. I think people are actually getting used to this, but it could be a problem with just the speed of change.
We are actually getting wealthier. US personal income per capita is actually 10 times better than it was a hundred years ago. People think, oh, gee, I’d like to live in 1900, but they don’t realize how poor we were and how little help we got from health technology, for example. So things were very, very poor a hundred years ago, and that’s going to continue. But we’re going to have to change our expectations as to what kind of jobs we can have and so on, so this is going to be very rapid change.
And we’ve seen this in the last two years. I mean, this AI revolution started two years ago, but you compare the technology two years ago to today. It’s quite astounding how much progress has been made.
IRA FLATOW: Well, do you think our job training can keep up with this as people will need to change the kinds of jobs and the education they get?
RAY KURZWEIL: Yes, and it’s not exactly clear what to do. I think we will actually have some government programs so that your baseline– I mean, some people can’t work, for example. There’s not going to be zero, and we’re going to move towards that. But then the real concern is having a job that gives you satisfaction and personal reflection and so on, not just being able to make a living. So that’s a nice problem to have, but that’s actually going to take a lot of concern on people who try to plan things like job training and so on.
IRA FLATOW: Ray, we have run out of time. It’s always a pleasure to talk to you, and I wish you good luck with your book. It’s a great read.
RAY KURZWEIL: Yes, we’re very much looking forward to it. It’s coming out on the 25th of this month.
IRA FLATOW: Ray Kurzweil, inventor, futurist, and author of The Singularity Is Nearer– When We Merge with AI. And if you want to read an excerpt of the book, you can do so on our website, sciencefriday.com/singularity.
Copyright © 2024 Science Friday Initiative. All rights reserved. Science Friday transcripts are produced on a tight deadline by 3Play Media. Fidelity to the original aired/published audio or video file might vary, and text might be updated or amended in the future. For the authoritative record of Science Friday’s programming, please visit the original aired/published recording. For terms of use and more information, visit our policies pages athttp://www.sciencefriday.com/about/policies/
