INTRODUCTION

“The period of time occupied by organic intelligence is just a thin sliver between early life and the long era of the machines.”

—Martin Rees, The Conversation, April 2017 interview

AI and Beyond

“A lot of cutting edge AI has filtered into general applications, often without being called AI, because once something becomes useful enough and common enough, it’s not labeled AI anymore.”

—Nick Bostrom, “AI Set to Exceed Human Brain Power,” CNN.com, 2006

Throughout history, the mysteries of the mind, the nature of thought, and the possibility of artificial beings have captivated artists, scientists, philosophers, and even theologians. Symbols and stories involving automata—moving mechanical devices made in imitation of living beings—permeate myth, art, music, and literature. Our fascination with artificial intelligence (AI)—apparently intelligent behavior by machines—is also reflected in the spooky or transcendent themes of blockbuster films or video games involving emotional robots and advanced intelligences that we can barely comprehend.

In this book, we’ll embark on a vast chronological journey from ancient games to advanced modern computing approaches involving artificial neural networks that learn and improve their performance, often with little or no task-specific programming and rules. Along the way, we’ll encounter odd and perplexing marvels like the mysterious copper knights of Arthurian legend. We’ll also encounter French inventor Jacques de Vaucanson’s Canard Digérateur, a hyper-realistic duck automaton that inspired American author Thomas Pynchon’s historical novel Mason & Dixon more than 250 years later, and the thirteenth-century Catalan philosopher Ramon Llull, who was among the first to explore a systematic approach to artificially generating ideas using a mechanical device. Skip forward to 1893, and we encounter the quirky and entertaining Electric Bob’s Big Black Ostrich, a story that—along with the series The Steam Man of the Prairies—is notable for reflecting the growing fervor for all things mechanical during the Victorian-era steampunk movement.

In more recent times, we encounter Arthur Samuel of IBM, who implemented one of the earliest computer programs for playing checkers in 1952, followed in 1955 by a program that learned to play the game without outside interference. Today, the term artificial intelligence often refers to systems designed to learn, solve problems, and interact with humans using natural-language processing. Intelligent personal assistants like Amazon’s Alexa, Apple’s Siri, and Microsoft’s Cortana all reflect some aspect of AI.

In this book, we’ll also address fascinating issues concerning the ethical use of AI and even the challenges of placing advanced AI entities, should they become dangerously super-intelligent, into a “leakproof” box in order to isolate them from the outside world. Of course, the boundaries and scope of AI morph through time, and some experts suggest inclusive definitions that admit a range of technologies that have helped human beings perform cognitive tasks. Thus, to give a richer understanding of AI history, I have also included several devices or machines that have provided answers to problems that typically require human thinking and human computation, including the abacus, the Antikythera Mechanism (125 BCE), ENIAC (1946), and more. After all, without these early technologies, we would not have the advanced chess-playing and vehicle-driving systems that exist in our modern world.

As you read this book, remember that even if we consider some of the historical ideas or predictions concerning artificial beings to be far-fetched, old ideas may become suddenly viable when implemented on faster, more advanced computer hardware. Our technical predictions—and even our myths—are, at a minimum, fascinating models of human understanding and creativity—and of how we reach across cultures and time to understand one another and learn about what we hold sacred or beneficial for society. However, even as we celebrate human imagination and ingenuity, it is vital to discuss unintended consequences, including the possible dangers of AI. As theoretical physicist Stephen Hawking told the BBC in a 2014 interview, “The development of full artificial intelligence could spell the end of the human race. . . . It would take off on its own, and redesign itself at an ever-increasing rate.” In other words, there’s a chance that AI entities will become so smart and capable that they will be able to continually improve themselves, creating a kind of superintelligence that could pose a great risk to humanity. This form of runaway technology growth, sometimes referred to as the technological singularity, could result in unimaginable changes to civilization, society, and human life.

So while the potential benefits of AI are numerous—self-driving cars, efficient business processes, and even companionship in countless arenas—humanity will need to be particularly cautious when developing autonomous weapons systems and over-relying on AI technologies with sometimes inscrutable mechanisms. For example, studies show the ease with which some AI (neural net) imaging systems can be “tricked” into misidentifying animals as rifles, or a picture of a plane as a dog, by altering images in ways that humans cannot perceive. If a terrorist can make a mall or a hospital look like a military target to a drone, the consequences could be dire. On the other hand, perhaps armed machines with appropriate sensors and rules of ethics could also reduce civilian casualties. Informed policy-making is needed to ensure that the potential dangers of AI entities do not overshadow their amazing benefits.

As we increasingly put our trust in AIs with many complex deep learning neural networks, one interesting area of research is developing AI systems that can explain to humans how they arrived at certain decisions. However, forcing AIs to explain themselves could potentially cripple them, at least in certain applications. Many of these machines can create far more intricate models of reality than humans could possibly understand. AI expert David Gunning has even suggested that the highest-performing system will be the least explainable.

Book Organization and Purpose

I have had a longtime fascination with computing and topics at the borderlands of science, and my goal in writing this book is to provide a wide audience with a brief guide to both curious and important practical ideas in the history of artificial intelligence, a term that wasn’t coined until 1955 by computer scientist John McCarthy. Each book entry is only a few paragraphs in length, allowing readers to jump in at any point in the book without having to sort through a lot of verbiage. Of course, this meant I couldn’t go into any depth on a subject. However, in the “Notes and References” section, there are suggestions for further reading and sourcing for various quotes or credentials of cited authors.

Touching on fields of study as diverse as philosophy, popular culture, computer science, sociology, and theology, the entries in this book also include topics that interested me personally. In fact, when I was younger, I became fascinated with Jasia Reichard’s Cybernetic Serendipity: The Computer and the Arts, published in 1969; the book featured computer-generated poetry, paintings, music, graphics, and more. I’m also particularly fascinated by the strides AI experts have made in the realm of art, using generative adversarial networks (GANs) to produce amazing photorealistic images of simulated faces, flowers, and birds. GANs make use of two neural networks pitted against each other—one network generating ideas and patterns, the other judging the results.

Today, the applications of AI seem limitless, and billions of dollars are invested in the development of AI each year. As you will discover, such technologies have been used to decipher the Vatican’s Secret Archives in an attempt to resolve the complicated handwritten texts in this huge historical collection. AI has also been used for predicting earthquakes, interpreting medical images and speech, and for predicting a person’s time of death based on patient information in a hospital’s electronic health records. AI has been deployed for generating jokes, mathematical theorems, US patents, games and puzzles, innovative designs for antennas, new paint colors, new fragrances, and more. Today, as many of us talk to our phones and other devices, our relationship with machines will continue to become even more intimate and humanlike in the future.

Organized chronologically according to the year associated with a key event, publication, or discovery, the dating of book entries is a question of judgment. Some dates are approximate; whenever possible, I tried to give a justification for the dates used.

You’ll also notice an increase in the number of entries after the year 1950. Daniel Crevier, author of AI: The Tumultuous History of the Search for Artificial Intelligence (1993), notes that in the 1960s, “AI blossomed in a thousand flowers. AI researchers applied their new programming techniques to many problems which, although real, had been carefully simplified, partly to isolate the problems to be addressed, but partly also to fit into the tiny memories of the computers available in those days.”

The mystery of consciousness, the limits of artificial intelligence, and the nature of the mind will be studied for years to come and have actually intrigued people from ancient times. The author Pamela McCorduck suggested in her book Machines Who Think that AI began with an ancient wish to “forge the gods.”

Future AI discoveries will be among humanity’s greatest achievements. The story of AI is not only about how we will shape our future but also about how humans will mesh with a landscape of accelerating intelligence and creativity all around us. What will it mean to be “human” a hundred years from now? What will society be like, given the increasing use of AI agents? How will jobs be affected? Will we fall in love with robots?

If AI methods and models are already being used to help determine who gets hired for a job, who we date, who makes parole, who is likely to develop a psychiatric disorder, and how to autonomously drive cars and drones, then how much control over our lives will we give to AIs of the future? As they increasingly make decisions for us, will AI units be easily fooled into making critical errors? How can AI researchers better understand why some machine-learning algorithms and architectures are more effective than others, while also making it easier for AI researchers to reproduce one another’s results and experiments?

Moreover, how can we ensure that AI-driven devices behave in an ethical manner, and will machines ever have mental states and feelings in the same sense that humans do? Surely, AI machines will help us think new thoughts and dream new dreams, functioning as prosthetics for our feeble brains. For me, AI cultivates a perpetual state of wonder about the limits of thought, the future of humanity, and our place in the vast space-time landscape that we call home.