Introduction
In "Deep Thinking," Garry Kasparov, one of the greatest chess players in history, takes us on a fascinating journey through the world of chess and artificial intelligence. The book explores the intersection of human intelligence and machine learning, using chess as a lens to examine the broader implications of technological advancement. Kasparov's unique perspective as both a chess grandmaster and a pioneer in human-computer competition provides valuable insights into the evolving relationship between humans and machines.
The Cultural Significance of Chess
Chess in the West: A Game for Nerds?
Kasparov begins by addressing the perception of chess in Western culture. Despite its ancient origins and centuries-long presence in Western society, chess has often been viewed as a game for intellectuals and eccentrics. This stereotype has been difficult to shake, even as Kasparov himself has attempted to challenge it through his public appearances and discussions on topics beyond chess.
The author notes that in Western schools, chess players often find themselves at the bottom of the social hierarchy. However, there are signs of change, particularly in the United States, where school chess programs are helping young children discover the game's appeal without preconceived notions.
Chess in Russia: A Revered Tradition
In contrast to the Western view, Kasparov describes the reverence for chess in Russia, where he grew up during the Soviet era. Chess was widely played and promoted, enjoying a status similar to popular sports like baseball in the United States. This tradition of respect for chess and chess players dates back to Tsarist times and was maintained even through the Russian Revolution and the Communist era.
The Soviet government went so far as to exempt elite chess players from military service during the Russian civil war, allowing them to participate in chess championships instead. This stark difference in cultural attitudes towards chess between Russia and the West highlights the game's unique position in different societies.
The Evolution of Chess-Playing Computers
Early Days: MANIAC 1 and Limited Capabilities
Kasparov traces the development of chess-playing computers from their humble beginnings in the 1950s. He describes MANIAC 1, developed in 1956 at a laboratory in Los Alamos, New Mexico, as one of the first computers with enough memory to store a chess program. Despite its impressive size (weighing about 1000 pounds), MANIAC 1's capabilities were limited. It could only play on a reduced board of 36 squares, without bishops, and lost to an experienced player even when the human opponent played without a queen.
However, MANIAC 1 did manage to beat a chess novice that same year, marking the first time artificial intelligence defeated a human in an intellectual game. This milestone, though modest, set the stage for rapid advancements in computer chess.
Moore's Law and Rapid Improvement
Kasparov explains how Moore's Law, which states that computer processing speeds double every two years, drove the rapid improvement in chess-playing computers. By 1977, computers could compete with the top 5% of human players, although they still made occasional game-losing errors.
A significant breakthrough came with the development of the alpha-beta algorithm in the 1970s. This algorithm allowed computers to automatically reject less effective moves, narrowing the number of possibilities they needed to evaluate. As a result, computers became faster at calculating possible moves and could even "think" several moves ahead.
The Impact of Computers on Employment
The Ongoing Debate: Humans vs. Machines
Kasparov addresses the broader implications of technological advancement, particularly the impact of computers on employment. He notes that debates pitting humans against machines date back to the Industrial Revolution when agricultural and manufacturing equipment began replacing human laborers.
The author traces this trend through different eras:
- The 1960s and 1970s: Precisely engineered machines made skilled laborers like watchmakers and laboratory assistants obsolete.
- The Information Revolution: The advent of the internet wiped out millions of service and support jobs, such as bank tellers and travel agents, replacing them with online e-services.
Kasparov predicts that even prestigious professions like doctors and lawyers may eventually be affected by technological advancements.
A Positive Perspective on Technological Progress
Despite the potential for job displacement, Kasparov argues against becoming sentimental over machines taking over human labor. He points out that technological progress has historically been beneficial for human civilization, leading to improvements in quality of life and advancements in human rights.
The author suggests that our ability to complain about the eradication of manual labor while enjoying the comforts of modern life is a sign of privilege. He emphasizes the need for adaptation, acknowledging that things won't return to the way they were. Instead, workers whose jobs have been replaced by artificial intelligence will need to be directed towards new types of technological and service jobs as they emerge.
The Rapid Development of Artificial Intelligence
From Simple Responses to Complex Questioning
Kasparov shares his experience of interacting with a robot named Artie at a robotics event in Oxford in 2016. He uses this anecdote to illustrate the rapid advancements in artificial intelligence, particularly in the realm of human-machine interaction.
The author explains that while computers have long been able to provide solutions, they were previously unable to formulate questions. However, this is no longer the case. Modern AI can ask questions, although it still struggles to determine which questions are truly important.
Kasparov describes how devices like Google Assistant and Amazon's Alexa work, using prompts and automated responses to create the illusion of authentic interaction. However, he notes that scientists are now working on enabling machines to formulate their own questions directly from harvested data, moving beyond the need for pre-programmed human prompts.
The Future of AI in Chess
The author explores how these advancements in AI might apply to chess. He explains that until recently, chess computers had strategies directly programmed into them, such as understanding the relative value of different pieces. Now, researchers are developing chess computers with only the most basic chess rules programmed, allowing them to work out everything else independently.
This approach could lead to completely novel strategies and plays that the computers could then teach to humans. Kasparov suggests that this development in chess AI is indicative of broader trends in artificial intelligence, where machines may surprise us not only with the data they produce but also with their methods of analysis and problem-solving.
The Psychological Nature of Chess for Humans
Chess as a Mental and Emotional Challenge
Kasparov delves into the psychological aspects of chess, comparing the mental exhaustion experienced after a chess match to the physical exhaustion felt after a track race. He argues that for humans, chess is ultimately a psychological game, where anxiety and mental fatigue can lead even the best players to make tactical mistakes.
The author references his own studies of famous grandmasters' matches, including his own, which revealed that psychological factors often play a crucial role in determining the outcome of a game. He highlights the approach of Emanuel Lasker, World Chess Champion from 1894 to 1921, who emphasized making moves that would make opponents uncomfortable, even if they weren't necessarily the most tactically sound.
The Emotionless Nature of Computer Chess
In contrast to the psychological complexity of human chess, Kasparov explains that computers approach the game purely as a matter of strategy. Computers are not affected by emotions or stress, allowing them to focus solely on calculating the most advantageous moves.
By 1985, computers were already capable of computing every possible combination of moves for the next three or four turns and selecting the most appropriate one. However, human players who could strategize at least five moves ahead still had an advantage. This comparison highlights the fundamental differences between human and computer approaches to chess, with humans relying on intuition and psychological factors, while computers depend on raw computational power.
The Power and Limitations of Data-Driven AI
Brute Force and Machine Learning
Kasparov discusses the importance of data in developing artificial intelligence, drawing parallels to Malcolm Gladwell's theory that success is more about practice than innate talent. For AI, the author argues, it's all about brute force – the ability to process vast amounts of data.
He cites the work of Donald Michie, a British researcher in artificial intelligence, who pioneered the use of large amounts of raw data to train computers. Michie's approach, first tested with tic-tac-toe in 1960, involved feeding the computer numerous examples of game moves and allowing it to derive basic principles from this data.
Kasparov draws a parallel between this approach and modern translation programs like Google Translate, which rely on millions of example sentences to piece together reasonable translations without actually understanding the languages involved.
The Potential for Errors in Data-Driven Systems
While acknowledging the power of data-driven AI, Kasparov also highlights its limitations. He recounts Michie's attempt to create a chess-playing machine in the 1980s by feeding it millions of chess moves from grandmaster games. The resulting computer became a great player but would occasionally make baffling moves, such as sacrificing its queen for no apparent reason.
This example illustrates how computers can misinterpret patterns in data, understanding certain moves or strategies in isolation without grasping the full context in which they should be applied. Kasparov uses this anecdote to emphasize that while data-driven AI can achieve impressive results, it can also lead to significant errors if not properly understood and implemented.
Learning to Lose: Kasparov's Experience with Chess Computers
The Emotional Challenge of Losing
Kasparov shares his personal experiences with losing, both against human opponents and computers. He admits that losing was never easy for him, often resulting in sleepless nights and even tantrums at award ceremonies. However, he argues that this intense dislike of losing is necessary for being a good competitor, as it drives one to continue improving and competing.
The author's impressive record – losing only 170 out of 2400 career matches against human opponents – underscores his competitive nature and skill. However, his encounters with chess computers presented a new challenge.
Facing Defeat by Computers
Kasparov recounts his first loss to a computer, Fritz 3, in a blitz chess tournament in Munich in May 1994. Although he won the overall tournament, this loss marked the first time a computer had defeated a world chess champion. This experience foreshadowed the more significant challenges to come.
The author then describes his famous matches against IBM's Deep Blue in 1996 and 1997. While he won the first match, Deep Blue emerged victorious in the rematch the following year. Kasparov explains that Deep Blue's ability to calculate an enormous number of possible moves for each turn ultimately gave it the edge.
This defeat was a turning point for Kasparov, forcing him to confront the reality that computers could now regularly beat him and would only become more powerful in the future. He describes this realization as a lesson in humility and acceptance of losing, highlighting the broader implications of AI surpassing human capabilities in specific domains.
The Persistence of Foul Play in Chess
Historical Examples of Gamesmanship
Kasparov delves into the less glamorous side of competitive chess, sharing anecdotes of foul play and gamesmanship. He recounts the bitter rivalry between Anatoly Karpov and Viktor Korchnoi during the 1978 World Championships in the Philippines, which led to some bizarre attempts at psychological warfare.
These incidents included Karpov hiring a psychologist to stare at Korchnoi throughout the match, supposedly to hypnotize or distract him. In response, Korchnoi recruited Indian sect members to meditate and stare at Karpov and his psychologist. The author also mentions the constant accusations of cheating between the two players, leading to investigations of various objects, including Korchnoi's chair and glasses, and famously, Karpov's yogurt.
Foul Play in the Age of Computer Chess
Kasparov explains that the introduction of computers in chess hasn't eliminated foul play; it has simply changed its form. He discusses the role of human intervention in computer chess matches, where technicians are allowed to perform certain tasks such as fixing bugs, restarting crashed computers, and adjusting computers' evaluative functions between games.
The author recalls his famous rematch with Deep Blue in 1997, during which the computer crashed twice and had to be restarted. He points out that such restarts could potentially be used to give computers an unfair advantage, as they erase the computer's memory tables and can lead to different move decisions. As a result, technicians' interventions are now more strictly regulated in computer chess competitions.
These examples serve to illustrate that while the nature of chess competition has changed with the introduction of computers, the human element – including the potential for foul play – remains a constant factor.
The Future of AI and Chess
Chess as a Milestone for AI
Kasparov reflects on the significance of chess in the development of artificial intelligence. He notes that chess, despite its complexity and beauty, ultimately proved simple enough for computers to master. The victory of Deep Blue over Kasparov in 1997, using only the processing power available at that time, marked a significant milestone in the field of AI.
New Frontiers for AI in Gaming
Looking to the future, Kasparov suggests that the next challenge for computer science will be to master more complex board games with many more squares and variables than chess. He specifically mentions the Chinese game Go as an example of a more complex challenge for AI.
This prediction has since been realized with the development of AI systems like AlphaGo, which have successfully mastered Go and other complex games. These advancements continue to push the boundaries of what's possible in artificial intelligence and machine learning.
Final Thoughts: The Ongoing AI Revolution
Kasparov concludes by emphasizing the rapid progress of artificial intelligence and its potential to surpass human intelligence in various domains. He notes that while current AI systems primarily rely on brute computing force and the ability to process vast amounts of data, a new revolution in artificial intelligence is on the horizon.
The author anticipates a future where computers can not only analyze data but also formulate questions from it and develop solutions independently of human input. This level of autonomy and creativity in AI would mark a truly new era in the relationship between humans and machines.
Kasparov's unique perspective as both a chess grandmaster and a pioneer in human-computer competition provides valuable insights into this evolving relationship. Through the lens of chess, he offers a thoughtful exploration of the challenges and opportunities presented by advancing artificial intelligence.
As we move forward, Kasparov's reflections remind us of the importance of adapting to technological change while also maintaining a critical eye on its implications. The story of chess and AI serves as a microcosm for broader societal shifts, encouraging us to think deeply about the future of human-machine interaction and the role we want AI to play in our lives.
In the end, "Deep Thinking" is not just a book about chess or artificial intelligence, but a meditation on human ingenuity, the nature of intelligence, and our ongoing quest to push the boundaries of what's possible. It challenges us to consider how we can harness the power of AI while preserving the uniquely human qualities that have driven our progress thus far.