Deep Thinking

INTRODUCTION

IT WAS A PLEASANT DAY in Hamburg on June 6, 1985, but chess players rarely get to enjoy the weather. I was inside a cramped auditorium, pacing around inside a circle of tables upon which rested thirty-two chessboards. Across from me at every board was an opponent, who moved promptly when I arrived at the board in what is known as a simultaneous exhibition. "Simuls," as they are known, have been a staple of chess for centuries, a way for amateurs to challenge a champion, but this one was unique. Each of my opponents, all thirty-two of them, was a computer.

I walked from one machine to the next, making my moves over a period of more than five hours. The four leading chess computer manufacturers had sent their top models, including eight bearing the "Kasparov" brand name from the electronics firm Saitek. One of the organizers warned me that playing against machines was different because they would never get tired or resign in dejection the way a human opponent would; they would play to the bitter end. But I relished this interesting new challenge—and the media attention it attracted. I was twenty-two years old, and by the end of the year I would become the youngest world chess champion in history. I was fearless, and, in this case, my confidence was fully justified.

It illustrates the state of computer chess at the time that it didn't come as much of a surprise, at least not in the chess world, when I achieved a perfect 32–0 score, winning every game, although there was one uncomfortable moment. At one point I realized that I was drifting into trouble in a game against one of the Kasparov models. If this machine scored a win or even a draw against me, people might suggest that I had thrown the game to get publicity for the company, so I had to intensify my efforts. Eventually I found a way to trick the machine with a sacrifice it should have refused and secure my clean sweep. From the human perspective, or at least from my perspective as the human in this equation, these were the good old days of human versus machine chess. But this golden age would be brutally short.

Twelve years later I was in New York City fighting for my chess life against just one machine, a $10 million IBM supercomputer nicknamed "Deep Blue." This battle, actually a rematch, became the most famous human-machine competition in history. Newsweek's cover called the it "The Brain's Last Stand" and a flurry of books compared it to Orville Wright's first flight and the moon landing. Hyperbole, of course, but not out of place at all in the history of our love-hate relationship with so-called intelligent machines.

Jump forward another twenty years to today, to 2017, and you can download any number of free chess apps for your phone that rival any human Grandmaster. You can easily imagine a robot in my place in Hamburg, circling inside the tables and defeating thirty-two of the world's best human players at the same time. The tables have turned, as they always do in our eternal race with our own technology.

Ironically, if a machine did perform a chess simul against a room full of human professional players, it would have more trouble moving from board to board and physically moving the pieces than it would have calculating the moves. Despite centuries of science fiction about automatons that look and move like people, and for all the physical labor today done by robots, it's fair to say that we have advanced further in duplicating human thought than human movement.

In what artificial intelligence and robotics experts call Moravec's paradox, in chess, as in so many things, what machines are good at is where humans are weak, and vice versa. In 1988, the roboticist Hans Moravec wrote, "It is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility." I wasn't aware of these theories at the time, and in 1988 it was safe to include checkers but not yet chess, but ten years later it was obviously the case in chess as well. Grandmasters excelled at recognizing patterns and strategic planning, both weaknesses in chess machines that, however, could calculate in seconds tactical complications that would take even the strongest humans days of study to work out.

This disparity gave me an idea for an experiment after my matches with Deep Blue attracted so much attention. You could also call it "if you can't beat 'em, join 'em," but I was eager to continue the computer chess experiment even if IBM was not. I wondered, what if instead of human versus machine we played as partners? My brainchild saw the light of day in a match in 1998 in León, Spain, and we called it Advanced Chess. Each player had a PC at hand running the chess software of his choice during the game. The idea was to create the highest level of chess ever played, a synthesis of the best of man and machine. It didn't quite go according to plan, as we'll see later, but the fascinating results of these "centaur" competitions convinced me that chess still had a lot to offer the worlds of human cognition and artificial intelligence.

In this belief I was hardly a pioneer; a chess-playing machine has been a holy grail since long before it was possible to make one. I just happened to be the human holding the grail when it was finally in science's grasp. I could run away from this new challenge or I could embrace it, which was really no choice at all. How could I resist? It was a chance to promote chess to a general audience beyond that reached even by Bobby Fischer's Cold War–era match against Boris Spassky and my own title duels with Anatoly Karpov. It had the potential to attract a new set of deep-pocketed sponsors to chess, especially tech companies. For example, Intel sponsored a Grand Prix cycle in the mid-1990s as well as my world championship match with Viswanathan Anand in 1995, played at the top of the World Trade Center. And then there was the irresistible curiosity I felt. Could these machines really play chess at the world championship level? Could they really think?

Humans have dreamed of intelligent machines since long before the technology to attempt one was conceived. In the late eighteenth century, a chess-playing mechanical automaton called the "Turk" was a wonder of the age. A carved wooden figure moved the pieces and, most remarkably, played a very strong game. Before it was destroyed in a fire in 1854, the Turk toured Europe and the Americas to great acclaim, claiming among its victims the famous chess aficionados Napoleon Bonaparte and Benjamin Franklin.

Of course it was a hoax; there was a human inside the cabinet under the table, hidden by an ingenious set of sliding panels and machinery. In another irony, today chess tournaments are plagued by cheaters who access super-strong computer programs to defeat their human opponents. Players have been caught using sophisticated signaling methods with accomplices, Bluetooth headsets in hats or electrical devices in shoes, and simply using a smartphone in the restroom.

The first real chess program actually predates the invention of the computer and was written by no less a luminary than Alan Turing, the British genius who cracked the Nazi Enigma code. In 1952, he processed a chess algorithm on slips of paper, playing the role of CPU himself, and this "paper machine" played a competent game. This connection went beyond Turing's personal interest in chess. Chess had a long-standing reputation as a unique nexus of the human intellect, and building a machine that could beat the world champion would mean building a truly intelligent machine.

Turing's name is forever attached to a thought experiment later made real, the "Turing test." The essence is whether or not a computer can fool a human into thinking it is human and if yes, it is said to have passed the Turing test. Even before I faced Deep Blue, computers were beginning to pass what we can call the "chess Turing test." They still played poorly and often made distinctively inhuman moves, but there were complete games between computers that wouldn't have looked out of place in any strong human tournament. As became clearer as the machines grew stronger every year, however, this taught us more about the limitations of chess than about artificial intelligence.

You cannot call the globally celebrated culmination of a forty-five-year-long quest an anticlimax, but it turned out that making a great chess-playing computer was not the same as making a thinking machine on par with the human mind, as Turing and others had dreamed. Deep Blue was intelligent the way your programmable alarm clock is intelligent. Not that losing to a $10 million alarm clock made me feel any better.

The AI crowd, too, was pleased with the result and the attention but dismayed by the fact that Deep Blue was hardly what their predecessors had imagined decades earlier when they dreamed of creating a machine to defeat the world chess champion. Instead of a computer that thought and played chess like a human, with human creativity and intuition, they got one that played like a machine, systematically evaluating up to 200 million possible moves on the chess board per second and winning with brute number-crunching force. This isn't to diminish the achievement in any way. It was a human achievement, after all, so while a human lost the match, humans also won.

After the unbearable tension of the match, exacerbated by IBM's questionable behavior and my suspicious human mind, I was in no mood to be a gracious loser. Not that I've ever been a good loser, I hasten to add. I believe accepting losses too easily is incompatible with being a great champion—certainly this was the case with me. I do believe in fighting a fair fight, however, and this is where I felt IBM had shortchanged me as well as the watching world.

Reexamining every aspect of that infamous match with Deep Blue for the first time in twenty years has been difficult, I admit. For two decades I have succeeded almost completely in avoiding and deflecting discussion about my Deep Blue matches beyond what was publicly known. There are many books about Deep Blue, but this is the first one that has all the facts and the only one that has my side of the story. Painful memories aside, it has also been a revealing and rewarding experience. My great teacher Mikhail Botvinnik, the sixth world champion, taught me always to seek the truth in the heart of every position. It has been fulfilling to finally find the truth at the heart of Deep Blue.

MY CAREER and my investigations into human-machine cognition did not end with Deep Blue, however; nor does this book. In fact, in both cases it's just the beginning. Competing head to head against a computer the way I did isn't the norm, although it was symbolic of how we are in a strange competition both with and against our own creations in more ways every day. My Advanced Chess experiment flourished online, where teams of humans and computers working together competed with remarkable results. Smarter computers are one key to success, but doing a smarter job of humans and machines working together turns out to be far more important.

These investigations led to visits to places like Google, Facebook, and Palantir, companies for whom algorithms are lifeblood. There have also been some more surprising invitations, including one from the headquarters of the world's largest hedge fund, where algorithms make or lose billions of dollars every day. There I met one of the creators of Watson, the Jeopardy-playing computer that could be called IBM's successor to Deep Blue. Another trip was to participate in a debate in front of an executive banking audience in Australia on what impact AI was likely to have on jobs in their industry. Their interests are quite different, but they all want to be on the cutting edge of the machine intelligence revolution, or at least to not be cut by it.

I've been speaking to business audiences for many years, usually on subjects like strategy and how to improve the decision-making process. But in recent years, I'm receiving more and more requests to talk about artificial intelligence and what I call the human-machine relationship. Along with sharing my thoughts, these appearances have given me the opportunity to listen closely to the interests of the business world regarding intelligent machines. Much of this book is dedicated to addressing these concerns and separating inevitable facts from conjecture and hyperbole.

In 2013, I was honored to become a senior visiting fellow at the Oxford Martin School, where I get to spend time with a constellation of brilliant expert minds. At Oxford, artificial intelligence is as much an area of philosophy as technology, and I enjoy trying to cross these streams. Their wonderfully named Future of Humanity Institute is the perfect place to collaborate on where the human-machine relationship is headed. My goal is to take some of the sophisticated, often arcane, expert research, predictions, and opinions and to serve as your translator and guide to their practical implications while adding my own insights and questions along the way.

I have spent most of my life thinking about how humans think and have found this to be an excellent basis for relating how machines think, and how they do not. In turn, this insight helps inform us as to what our machines can and cannot do… yet.

THE NINETEENTH-CENTURY African American folk legend of John Henry pits the "steel-driving man" in a race against a new invention, a steam-powered hammer, bashing a tunnel through a mountain of rock. It was my blessing and my curse to be the John Henry of chess and artificial intelligence, as chess computers went from laughably weak to nearly unbeatable during my twenty years as the world's top chess player.

As we will see, this is a pattern that has repeated over and over for centuries. People scoffed at every feeble attempt to substitute clumsy, fragile machines for the power of horses and oxen. We laughed at the idea that stiff wood and metal could replicate the soaring grace of the birds. Eventually we have had to concede that there is no physical labor that couldn't be replicated, or mechanically surpassed.

It is also now widely accepted that this inexorable advance is something to celebrate, not fear, although it is usually two steps forward and one step back in this regard. With every new encroachment of machines, the voices of panic and doubt are heard, and they are only getting louder today. This is partly due to the differences in what, and who, is being replaced. The horses and oxen couldn't write letters to the editor when cars and tractors came along. Unskilled laborers also lacked much of a voice, and were often considered lucky to be freed from their backbreaking toil.

So it went over the decades of the twentieth century, with countless jobs lost or transformed by automation. Entire professions disappeared with little time to mourn them. The elevator operators' union was seventeen thousand strong in 1920, although its ability to paralyze cities with strikes like the one its members staged in New York in September 1945 surely cost them more than a few mourners when automatic push-button elevators began to replace them in the 1950s. According to the Associated Press, "Thousands struggled up stairways that seemed endless, including the Empire State Building, tallest structure in the world."

Good riddance, you might imagine. But the worries about operatorless elevators were quite similar to the concerns we hear today about driverless cars. In fact, I learned something surprising when I was invited to speak to the Otis Elevator Company in Connecticut in 2006. The technology for automatic elevators had existed since 1900, but people were too uncomfortable to ride in one without an operator. It took the 1945 strike and a huge industry PR push to change people's minds, a process that is already repeating with driverless cars. The cycle of automation, fear, and eventual acceptance goes on.

Of course, what an observer calls freedom and disruption, a worker calls unemployment. The educated classes in the developed world have long had the luxury of lecturing their blue-collar brethren about the glories of the automated future. Service personnel have been on the block for decades—their friendly faces, human voices, and quick fingers replaced by ATMs, photocopiers, phone trees, and self-checkout lines. Airports have iPads instead of food servers. No sooner did massive call centers spring up around India than automated help-desk algorithms begin replacing them.

It is far easier to tell millions of newly redundant workers to "retrain for the information age" or to "join the creative entrepreneurial economy" than to be one of them or to actually do it. And who can say how quickly all that new training will also become worthless? What professions today can be called "computer proof"? Today another set of tables has turned, or rather, desks. The machines have finally come for the white collared, the college graduates, the decision makers. And it's about time.

JOHN HENRY won his race against the machine only to die on the spot, "his hammer in his hand." I was spared such a fate myself, and humans are still playing chess, in fact more today than ever before. The doomsayers who said no one would want to play a game that could be dominated by a computer have been proven wrong. This seems obvious, considering how we also still play far simpler games like tic-tac-toe and checkers, but doomsaying has always been a popular pastime when it comes to new technology.

I remain an optimist if only because I've never found much advantage in the alternatives. Artificial intelligence is on a path toward transforming every part of our lives in a way not seen since the creation of the Internet, perhaps even since we harnessed electricity. There are potential dangers with any powerful new technology and I won't shy away from discussing them. Eminent individuals from Stephen Hawking to Elon Musk have expressed their fear of AI as a potential existential threat to mankind. The experts are less prone to alarming statements, but they are quite worried too. If you program a machine, you know what it's capable of. If the machine is programming itself, who knows what it might do?

The airports with their self-check-in kiosks and restaurants full of iPads are staffed by thousands of human workers (most using machines) in the long security lines. Is it because they can do things no machine can do? Or, like operating an elevator and driving a car, is it because at first we don't trust machines to do a job where lives are at risk? Elevators became much safer as soon as the human operators were replaced. The human-hating Skynet from the Terminator movies could hardly do a better job of killing people than we do killing ourselves with cars. Human error is responsible for over 50 percent of plane crashes, although overall air travel is getting safer as it becomes more automated.

In other words, fail-safes are required, but so is courage. When I sat across from Deep Blue twenty years ago I sensed something new, something unsettling. Perhaps you will experience a similar feeling the first time you ride in a driverless car, or the first time your new computer boss issues an order at work. We must face these fears in order to get the most out of our technology and to get the most out of ourselves.

Many of the most promising jobs today didn't even exist twenty years ago, a trend that will continue and accelerate. Mobile app designer, 3D print engineer, drone pilot, social media manager, genetic counselor—to name just a few of the careers that have appeared in recent years. And while experts will always be in demand, more intelligent machines are continually lowering the bar to creating with new technology. This means less training and retraining for those whose jobs are taken by robots, a virtuous cycle of freeing us from routine work and empowering us to use new technology productively.

Machines that replace physical labor have allowed us to focus more on what makes us human: our minds. Intelligent machines will continue that process, taking over the more menial aspects of cognition and elevating our mental lives toward creativity, curiosity, beauty, and joy. These are what truly make us human, not any particular activity or skill like swinging a hammer—or even playing chess.