AI 2041: Ten Visions for Our Future Read online

  The stories in this book are a work of fiction. Names, characters, business, events, and incidents are the products of the authors’ imaginations. Any resemblance to actual persons, living or dead, or actual events is purely coincidental.

  Copyright © 2021 by Kai-Fu Lee and Chen Qiufan

  All rights reserved.

  Published in the United States by Currency, an imprint of Random House, a division of Penguin Random House LLC, New York.

  Currency and its colophon are trademarks of Penguin Random House LLC.

  Library of Congress Cataloging-in-Publication Data

  Names: Lee, Kai-Fu, author. | Chen, Qiufan, author.

  Title: AI 2041 / by Kai-Fu Lee and Chen Qiufan.

  Description: First edition. | New York: Currency, [2021] | Includes index.

  Identifiers: LCCN 2021012928 (print) | LCCN 2021012929 (ebook) | ISBN 9780593238295 (hardcover; acid-free paper) | ISBN 9780593238301 (ebook)

  Subjects: LCSH: Artificial intelligence in literature. | Artificial intelligence.

  Classification: LCC Q335 .L423 2021 (print) | LCC Q335 (ebook) | DDC 006.3—dc23

  LC record available at https://lccn.loc.gov/​2021012928

  LC ebook record available at https://lccn.loc.gov/​2021012929

  International edition ISBN 9780593240717

  Ebook ISBN 9780593238301

  crownpublishing.com

  Book design by Edwin Vazquez, adapted for ebook

  Cover Design: Will Staehle

  ep_prh_5.7.1_c0_r0

  Contents

  Cover

  Title Page

  Copyright

  Epigraph

  Introduction by Kai-Fu Lee: The Real Story of AI

  Introduction by Chen Qiufan: How We Can Learn to Stop Worrying and Embrace the Future with Imagination

  Chapter One: The Golden Elephant

  Analysis: Deep Learning, Big Data, Internet/Finance Applications, AI Externalities

  Chapter Two: Gods Behind the Masks

  Analysis: Computer Vision, Convolutional Neural Networks, Deepfakes, Generative Adversarial Networks (GANs), Biometrics, AI Security

  Chapter Three: Twin Sparrows

  Analysis: Natural Language Processing, Self-Supervised Training, GPT-3, AGI and Consciousness, AI Education

  Chapter Four: Contactless Love

  Analysis: AI Healthcare, AlphaFold, Robotic Applications, COVID Automation Acceleration

  Chapter Five: My Haunting Idol

  Analysis: Virtual Reality (VR), Augmented Reality (AR), and Mixed Reality (MR), Brain-Computer Interface (BCI), Ethical and Societal Issues

  Chapter Six: The Holy Driver

  Analysis: Autonomous Vehicles, Full Autonomy and Smart Cities, Ethical and Social Issues

  Chapter Seven: Quantum Genocide

  Analysis: Quantum Computers, Bitcoin Security, Autonomous Weapons and Existential Threat

  Chapter Eight: The Job Savior

  Analysis: AI Job Displacement, Universal Basic Income (UBI), What AI Cannot Do, 3Rs as a Solution to Displacement

  Chapter Nine: Isle of Happiness

  Analysis: AI and Happiness, General Data Protection Regulation (GDPR), Personal Data, Privacy Computing Using Federated Learning and Trusted Execution Environment (TEE)

  Chapter Ten: Dreaming of Plenitude

  Analysis: Plenitude, New Economic Models, the Future of Money, Singularity

  Acknowledgments

  Other Titles

  About the Authors

  What we want is a machine that can learn from experience.

  —Alan Turing

  Any sufficiently advanced technology is indistinguishable from magic.

  —Arthur C. Clarke

  INTRODUCTION BY KAI-FU LEE

  THE REAL STORY OF AI

  Artificial intelligence (AI) is smart software and hardware capable of performing tasks that typically require human intelligence. AI is the elucidation of the human learning process, the quantification of the human thinking process, the explication of human behavior, and the understanding of what makes intelligence possible. It is mankind’s final step in the journey to understanding ourselves, and I hope to take part in this new, but promising science.

  I WROTE THESE words as a starry-eyed student applying to Carnegie Mellon University’s PhD program almost forty years ago. Computer scientist John McCarthy coined the term “artificial intelligence” even earlier—at the legendary Dartmouth Summer Research Project on Artificial Intelligence in the summer of 1956. To many people, AI seems like the quintessential twenty-first-century technology, but some of us were thinking about it decades ago. In the first three and a half decades of my AI journey, artificial intelligence as a field of inquiry was essentially confined to academia, with few successful commercial adaptations.

  AI’s practical applications once evolved slowly. In the past five years, however, AI has become the world’s hottest technology. A stunning turning point came in 2016 when AlphaGo, a machine built by DeepMind engineers, defeated Lee Sedol in a five-round Go contest known as the Google DeepMind Challenge Match. Go is a board game more complex than chess by one million trillion trillion trillion trillion times. Also, in contrast to chess, the game of Go is believed by its millions of enthusiastic fans to require true intelligence, wisdom, and Zen-like intellectual refinement. People were shocked that the AI competitor vanquished the human champion.

  AlphaGo, like most of the commercial breakthroughs in AI, was built on deep learning, a technology that draws on large data sets to teach itself things. Deep learning was invented many years ago, but only recently has there been enough computing power to demonstrate its efficacy, and sufficient training data to achieve exceptional results. Compared to when I made my cold start in AI forty years ago, we now have about one trillion times more computing power available for AI experimentation, and storing the necessary data is fifteen million times cheaper. The applications for deep learning—and its related AI technologies—will touch nearly every aspect of our lives.

  AI is now at a tipping point. It has left the ivory tower. The days of slow progress are over.

  In just the past five years, AI has beaten human champions in Go, poker, and the video game Dota 2, and has become so powerful that it learns chess in four hours and plays invincibly against humans. But it’s not just games that it excels at. In 2020, AI solved a fifty-year-old riddle of biology called protein folding. The technology has surpassed humans in speech and object recognition, served up “digital humans” with uncanny realism in both appearance and speech, and earned passing marks on college entrance and medical licensing exams. AI is outperforming judges in fair and consistent sentencing, and radiologists in diagnosing lung cancer, as well as powering drones that will change the future of delivery, agriculture, and warfare. Finally, AI is enabling autonomous vehicles that drive more safely on the highway than humans.

  As AI continues to advance and new applications blossom, where does it all lead?

  In my 2018 book AI Superpowers: China, Silicon Valley, and the New World Order, I addressed the proliferation of data, the “new oil” that powers AI. The United States and China are leading the AI revolution, with the United States leading research advances and China more swiftly tapping big data to introduce applications for its large population. In AI Superpowers, I predicted new advances, from big-data-driven decisio
n-making to machine perception to autonomous robots and vehicles. I projected that AI’s new applications in digital industries, finance, retail, and transportation would build unprecedented economic value, but also create problems related to the loss of human jobs and other issues. AI is an omni-use technology that will penetrate virtually all industries. Its effects are being felt in four waves, beginning with Internet applications, followed by applications in business (e.g. financial services), perception (think smart cities), and autonomous applications, like vehicles.

  Four waves of AI applications are disrupting virtually all industries.

  By the time you read this new book in late 2021 or beyond, the predictions I made in AI Superpowers will have largely become reality. We must now look ahead to new frontiers. As I’ve traveled the world talking about AI, I’m constantly asked, “What’s next?” What will happen in another five, ten, or twenty years? What will the future hold for us humans?

  These are essential questions for our moment in history, and everyone working in the technology space has an opinion. Some believe that we’re in the midst of an “AI bubble” that will eventually pop, or at least cool off. Those with more drastic and dystopian views believe everything from the notion that AI giants will “hijack our minds” and form a utopian new race of “human cyborgs” to the arrival of an AI-driven apocalypse. These various predictions may be born out of genuine curiosity or understandable fear, but they are usually speculative or exaggerated. They miss the complete picture.

  Speculation varies wildly because AI appears complex and opaque. I’ve observed that people often rely on three sources to learn about it: science fiction, news, and influential people. In science fiction books and TV shows, people see depictions of robots that want to control or outsmart humans, and superintelligence turned evil. Media reports tend to focus on negative, outlying examples rather than quotidian incremental advances: autonomous vehicles killing pedestrians, technology companies using AI to influence elections, and people using AI to disseminate misinformation and deepfakes. Relying on “thought leaders” ought to be the best option, but unfortunately most who claim the title are experts in business, physics, or politics, not AI technology. Their predictions often lack scientific rigor. What makes things worse is that journalists tend to quote these leaders out of context to attract eyeballs. So, it is no wonder that the general view about AI—informed by half-truths—has turned cautious and even negative.

  To be sure, aspects of AI development deserve our scrutiny and caution, but it is important to balance these concerns with exposure to the full picture and potential of this crucially important technology. AI, like most technologies, is inherently neither good nor evil. And like most technologies, AI will eventually produce more positive than negative impacts on our society. Think about the tremendous benefits of electricity, mobile phones, and the Internet. In the course of human history, we have often been fearful of new technologies that seem poised to change the status quo. In time, these fears usually go away, and these technologies become woven into the fabric of our lives and improve our standard of living.

  I believe there are many exciting applications and scenarios in which AI can profoundly enhance our society. Firstly, AI will create tremendous value to our society—PricewaterhouseCoopers estimates $15.7 trillion by 2030—which will help reduce hunger and poverty. AI will also create efficient services that will give us back our most valuable resource—time. It will take over routine tasks and liberate us to do more stimulating or challenging jobs. Lastly, humans will work symbiotically with AI, with AI performing quantitative analysis, optimization, and routine work, while we humans contribute our creativity, critical thinking, and passion. Each human’s productivity will be amplified, allowing us to realize our potential. The profound contributions AI is poised to make to humanity need to be explored as deeply as its challenges.

  Amid what seems like a feedback loop of negative stories about AI, I believe it’s important to tell these other stories, too, and answer that question of “What happens next?” So I decided to write another book about AI. This time, I wanted to extend the horizon a bit further—to imagine the future of the world and our society in twenty years’ time, or 2041. My aim is to tell the “real” AI story, in a way that is candid and balanced, but also constructive and hopeful. This book is based on realistic AI, or technologies that either already exist or can be reasonably expected to mature within the next twenty years. These stories offer a portrait of our world in 2041, based on technologies with a greater than 80-percent likelihood of coming to pass in that timeframe. I may overestimate or underestimate some. But I believe this book represents a responsible and likely set of scenarios.

  How can I be so confident? Over the past forty years, I have been involved in AI research and product development at Apple, Microsoft, and Google, and managed $3 billion in technology investments. So I have hands-on experience with the time and processes needed to take a technology from academic paper to pervasive product. Further, as an adviser to governments on AI strategy, I can make predictions based on my knowledge of policy and regulation frameworks, and the reasoning behind them. Also, I avoid making speculative predictions about fundamental breakthroughs and rely mostly on applying and extrapolating the future of existing technologies. Since AI has penetrated less than 10 percent of our industries, there are many opportunities to reimagine our future with AI infusion into these fields. In short, I believe that even with few or no breakthroughs, AI is still poised to make a profound impact on our society. And this book is my testimony.

  I’ve been told that one of the reasons that AI Superpowers made an impact on readers was that it was accessible to people with no prior knowledge of AI. So when I embarked on this new book, I asked: What can I do to tell stories about AI in a way that makes them even more widely appealing? The answer, of course, was to work with a good storyteller! I decided to reach out to my former Google colleague Chen Qiufan. After Google, I started a venture capital firm. Qiufan did something more adventurous—he became an award-winning science fiction writer. I was delighted that Qiufan agreed to work with me on the project, and to dovetail his creativity with my judgment on what technology will be capable of in twenty years. We both believed that imagining the feasible technologies within a twenty-year period and embedding them in stories would be quite engaging, and we wouldn’t even have to resort to teleportation or aliens to mesmerize our readers.

  Qiufan and I worked out a unique arrangement. I first created a “technology map” that projected when certain technologies would mature, how long it would take to gather data and iterate AI, and how easy it would be to build a product in various industries. I also accounted for possible externalities—challenges, regulations, and other deterrents, as well as story-worthy conflicts and dilemmas that might emerge alongside these technologies. With my input on the technological components, Qiufan then flexed his talents—dreaming up the characters, settings, and plotlines that would bring these themes to life. We worked to make each story engaging, provocative, and technologically accurate. After each one, I offer my technology analysis, digging into the forms of AI revealed and their implications for human life and society. We organized the stories to cover all key aspects of AI, and roughly ordered them from basic to advanced technologies. The sum of these parts, we hope, is a uniquely engaging and accessible primer on AI.

  We named our book AI 2041 because that is twenty years from the initial publication of this book. But it didn’t slip our notice that the digits “41” happen to look a bit like “AI.”

  Many of our readers may love the wonderful storytelling of science fiction, but I imagine there are others who may not have picked up a novel or a collection of short stories since college. That’s okay. If you fall into that camp, think of AI 2041 not as “science fiction” but as “scientific fiction.” The stories are set in wide-ranging locations around the world. In some, you may recognize a world that see
ms not too dissimilar from your own—with narratives that draw on existing customs and habits, albeit with an AI twist. In others, AI has transformed human life dramatically. Both AI enthusiasts and skeptics will have plenty to think about. Creating a book with a significant fiction component is inherently riskier than writing a nonfiction book that simply describes the present and asks questions about the future. Qiufan and I sought to be bold with our narratives, and we believe the stories that follow will strike a chord with every open-minded reader whose imagination is large enough to ponder what the future holds.

  The first seven stories were designed to cover technology applications for different industries in increasing technological complexity, along with their ethical and societal implications. The last three stories (plus chapter 6, “The Holy Driver”) focus more on social and geopolitical issues raised by AI, such as the loss of traditional jobs, an unprecedented abundance of goods, exacerbated inequality, an autonomous weapons arms race, trade-offs between privacy and happiness, and the human pursuit of a higher purpose. These are profound changes, and humans may embrace them with compassion, exploit them with malice, capitulate to them with resignation, or be inspired by them to reinvent ourselves. In the final four stories, we decided to show four possible variations and different pathways, as a way of underscoring that the future is not yet written.

  We hope the stories entertain you while deepening your understanding of AI and the challenges it poses. We also hope that the book’s road map of the coming decades will help you prepare yourself to capture the opportunities and confront the challenges that the future will bring. Most of all, we hope you will agree that the tales in AI 2041 reinforce our belief in human agency—that we are the masters of our fate, and no technological revolution will ever change that.