 Everyone is talking about the deep learning revolution. Everyone is saying that artificial intelligence will transform our society. But I'd like to add one more thought. AI is transforming the way that we do science. It's often said that seeing is believing. And sometimes, science is just the simple act of seeing. Something incredible. Through his telescope, Galileo saw the moons of Jupiter. And Leuvenuck, looking through his microscope, saw microbial life forms for the very first time. So let's look at the brain, magnified by 100,000 times. You can think of these as the pages of a 3D coloring book. It would have taken a century for a person to color all these neurons, even with no coffee breaks or sleep. Fortunately, an AI did the coloring for us automatically. If you click on a 2D cross-section of any neuron, you see the entire 3D shape now. You see cell bodies, you see branches. Dendrites studded with thorns, like long-stemmed roses. And if we zoom out, we can see what you learned in high school science class. Neurons resemble trees. But don't get the wrong impression. Don't think that neurons are floating in empty space. Actually, the brain is packed, full of neural branches, entangled in complex forms. And when two neurons touch, we can sometimes see a synapse. Here, at this synapse, a purple neuron sends signals to a yellow neuron. Why does neuroscience need AI? Automated microscopes now produce images, torrents of images, so enormous that no human can comprehend them. We have stretched the notion of science as seen to its limit. Today, we need AI to do the seeing for us. Just one cubic millimeter of brain contains a billion synapses and five kilometers of neural branches. If the AI can see all that, it can tell us exactly how these neurons are wired to each other. It can give us the ultimate brain map, the connectome. And there's more information. Neurons blink on and off in this green video, shot through a microscope looking at the living, functioning brain of an animal. And overlaid on this activity map, we see the electron microscopic images of the same neurons. So the map includes both structure and function. And you've been looking at the neocortex. It's the size of a dish towel. It's crumpled up to fit inside your skull. It's huge in humans, but small in mice and monkeys. And many neuroscientists think that this structure of the brain contains the secrets of human intelligence. By studying its connectome, we hope to understand how much genes determine the connectome. And we also, by comparing with the activity map, hope to understand how the connectome is shaped by learning from experiences. So neuroscientists think that every time you have an experience, it alters your connectome. And that's how memories are stored. And that's why I coined a slogan, I am my connectome. When will we be able to read memories from connectomes? Well, the first steps are being taken today. So we like to think that we are smarter than animals, which is mysterious because the neocortex of all animal species, all mammalian species, looks superficially the same. Are we smarter just because we have more neocortex? Our brains are bigger? Or is it because there's some difference in the connectional organization? 1960s-era neuroscience led to artificial neural nets, which led to deep learning as we know it today. And we're going to have a radically new understanding of the neocortex based on the new kinds of data we receive. That's going to change our knowledge of human intelligence and maybe inspire a new generation of AI. Now, a cubic millimeter of brain amounts to one petabyte of data. Your brain is one million times larger, and that's a zettabyte of data. If we're able to map your connectome 20 years from now, will we be able to simulate your mind? Now, the pioneers of AI, way back when, took inspiration from the brain to build artificial neural nets, and now the tables are turned, AI has grown up. AI is being used to accelerate progress in understanding the brain. And so we're closing that feedback loop. It's going to be very powerful when we do. Philosophers love to ask the question, is the brain complex enough to understand itself? Whoa. My answer is that humans, assisted by artificial intelligence, will conquer the final frontier, our brains. Thanks very much.