 The singularity is not a single soundbite as a concept of the number of underlying ideas. One is that information technology grows exponentially. And it turns out to be amazingly predictable. I mean, people say you can't predict the future. It turns out the exponential progression of information technology, and it's not just computation. It's things like genetic sequencing, brain scanning. Any way you can measure the underlying information properties grows at an exponential pace. And amazingly predictable. I have a graph of the price performance of computers going back to 1890, the first automated American census. It's a very smooth doubly exponential curve through 110 years, completely unperturbed by little things like two world wars, the Cold War, the Great Depression, and all the other history that we experienced. Genetic sequencing has doubled every year since the project was announced. And exponential growth is surprising because you start out doubling little numbers. It looks like not much is happening. And then it becomes quite explosive. If I take 30 steps linearly, 1, 2, 3, 4, 5, I get to 30. If I take 30 steps exponentially, 2, 4, 8, 16, I get to a billion. Ultimately, it makes a huge difference. Intelligence inherently predicts the future, but our built-in intelligence, what's hardwired, is linear thinking. We think in a linear manner. That worked very well 1,000 years ago and we saw something coming at us through the corner of our eye and we would make a linear prediction about where that animal would be. And that served our purposes quite well. And that intuition has survived to the present time. And even very sophisticated thinkers, when they're thinking about the future, let's say like a brain scientist who I debated recently who's studying one ion channel in a dendrite and did the mental experiment of, well, how long would it take me to reverse engineer the whole brain using today's tools and today's supercomputers and today's scanners? And he came up with the conclusion this would be centuries. And what he failed to take into consideration is that the tools, the scanning, the computers, everything is doubling in power every year. These tools will be 1,000 times more powerful in 10 years. So exponential growth of information technology is not just Moore's law. It's not just computers. It's anywhere we can measure the information technology. And ultimately that's quite explosive. And this is not just looking forward because I've been making these predictions for 30 years. When I was a student at MIT, we all shared one computer, took up half a building. The computer in your cell phone today is a million times more affordable. It's a million times cheaper, 1,000 times more powerful. That's a billion-fold increase in price performance that we've already experienced. It's 100,000 times smaller. And what used to take up a building now fits in your pocket. What fits in your pocket today will fit inside a blood cell 25 years from now. And the price performance will again be a billion times more powerful because even the rate of exponential growth is speeding up. So that's the first idea underlying the singularity. The second is that we will capture human intelligence in a machine. And I've been very consistent that the date for that is 2029. Long before that, we will have computers that are powerful enough to simulate all the regions of the human brain at very low cost. And that's pretty much a mainstream view now. What is perhaps more controversial is, okay, we'll have the hardware, but what about the software? Isn't that stuck in the mud? And I go to extensive lengths in the singularities here to point out that we're progressing at an exponential pace. 10, 15 years ago, many artificial intelligence applications were just experiments. Today, they permeate our economic infrastructure. Every time you send an email or connect the cell phone call intelligent algorithm through the information, pick up any product that was designed with intelligent computer assisted design software, inventory levels controlled by intelligent just-in-time inventory systems assembled in robotic factories, get an electrocardiogram and comes back with an automated diagnosis that rivals that of doctors, same-for-bud-cell images, intelligent algorithms, flying land airplanes, guide intelligent weapons systems automatically detect credit card fraud. And I can mention a hundred other applications of intelligent software doing tasks that used to be done by human intelligence and are now doing them just as well by artificial intelligence. And you can say, okay, well, these are narrow examples of AI and it's narrow AI, but the narrowness is gradually getting broader. And what's underlying this is that we are making exponential progress in understanding the best example we have of human intelligence, which is the human brain itself and it's not hidden from us. And there again, we're making very smooth, very predictable exponential progress. The spatial resolution of brain scanning is doubling every year. The amount of data we're getting is doubling every year. We're showing that we can turn this data into working models and simulations of brain regions. There's already 20 regions of the brain that have been modeled and simulated, like the auditory cortex, the visual cortex, the cerebellum where we do our scale formation, like catching a fly ball, slices of the cerebral cortex where we do our abstract reasoning, and the scale, the sophistication, the detail of these simulations is scaling up at an exponential pace. In chapter 4 of Singularity is Near that we will have the models and simulations of all the regions of the brain within 20 years. And that will provide the software for human intelligence. Once we capture human level intelligence in a machine, that will be in a very important threshold. First of all, it's not, in my view, just artificial intelligence to compete with us and displace us. We are going to use it to extend and expand who we are. We are going to leverage with this technology. It's already in our pockets. Ultimately, it will make its way into our bodies and brains. If I were to say, well, someday you'll have millions of intelligence, blood cell-sized devices in your bloodstream keeping you healthy from inside, going inside your brain and interacting with your brain, putting you on the internet and expanding your intelligence. You'd say, well, that sounds kind of futuristic. And I'd point out this already dozens of experiments with blood cell-sized devices that are keeping animals healthy in experiments. One scientist cured type 1 diabetes in rats with a blood cell-sized device, 7 nanometer pores, let's insulin out in a controlled fashion. At MIT, they have a blood cell-sized device that can detect cancer cells based on the antigens on their surface, latch onto them and destroy them, and there are dozens of other experiments like that. You can put a computer in your brain today and it will replace the biological neurons destroyed by that disease. The latest generation now allows you to download new software to the computer inside your brain from outside the patient, and that's an FDA-approved therapy. If you look at what we can do today, that neural implant is not blood cell size, it's pea size, but it will be 100,000 times smaller in 25 years and will be the size of a blood cell and will be a lot more sophisticated. So once we get to that threshold, artificial intelligence, working together with our biological intelligence, will be able to solve problems, design even more intelligent AI, and the law of accelerating returns will continue to make the hardware devices more powerful. When you get to the 2040s, the portion of our civilization's intelligence that's amplified by artificial intelligence will be a billion times greater than the biological intelligence that gave rise to all of this. In other words, the human civilization will expand its intelligence a billion fold because of the ongoing exponential explosion of intelligence from developing the artificial form of it. And that's such a profound transformation that it's hard to see past the horizon of that. So we use a metaphor from physics where singularity in physics of black hole has an event horizon around it that's hard to see beyond because none of the information can easily get out in the classical sense. So, however, in physics we can use our intelligence to talk about what life is like beyond the event horizon of a black hole, even though we can't easily see beyond it. And we can similarly do that today, talk about what life will be like past this event horizon even though it's profoundly transformative. And what we can say about it is we're going to amplify our intelligence. And intelligence is the facility we have that enables us to solve problems. All the major problems we have is the society, whether it's the environment or energy or health or disease. We use our intelligence to overcome those problems. And we already use our machine intelligence to help us with that. There isn't any scientific engineering and mathematical work, even work in the arts like music and graphic art that isn't amplified by using our computerized tools. And as these tools get more intelligent, more powerful we're going to continue to expand our reach. That's what human beings do. And ultimately the singularity represents a billion fold increase in the intelligence we have to apply to these problems.