Alberto's Blog
1 / 0

Creativity vs AI

The evolution of creativity in the game in the era of artificial intelligence

The Challenge

Artificial Intelligence has radically changed the game of chess, introducing new ideas and strategies. With the arrival of AlphaZero, a paradigm shift occurred: engines no longer just calculate millions of moves per second through brute force, but learn from their own experiences.

The Challenge

From the mid-twentieth century to today, the relationship between chess and artificial intelligence has gone through numerous historic moments. Among these, we will explore two fundamental milestones: the legendary challenge between Kasparov and Deep Blue, the match that marked an era, and the arrival of AlphaZero, which revolutionized the way of conceiving the game.

The Match of the Century

Kasparov’s challenge against Deep Blue, symbol of human creativity against computational power.

AlphaZero Against Stockfish

The revolution of AlphaZero against Stockfish v8, where AlphaZero learned to play from scratch through self-learning.

How Do Engines Work?

An engine is a program that analyzes positions and calculates the best possible moves. Traditional ones use complex evaluation functions and advanced search algorithms to find the best move.

Their power depends on hardware: Deep Blue used specialized chess chips, while modern engines like Stockfish exploit the power of multi-core CPUs to calculate millions of positions per second.

Classic engines like Deep Blue and Stockfish v8 combine deep search with evaluation functions based on parameters programmed by humans, exploring enormous quantities of moves before making a decision.

However, they had some limitations. While calculating very deeply, they struggled to understand long-term strategic plans that require intuition beyond pure calculation. Additionally, they had difficulty evaluating positional sacrifices with intangible compensations like initiative or piece coordination. They were constrained by fixed parameters set by programmers.

With the arrival of AlphaZero, the paradigm changed radically: for the first time, an engine learned on its own by playing millions of games against itself, developing deep strategic understanding without being programmed with predefined rules.

The Match of the Century

1996 Match

#1#2#3#4#5#6Final
Kasparov01½½114
Deep Blue10½½002

1997 Rematch

#1#2#3#4#5#6Final
Kasparov10½½½02½
Deep Blue01½½½13½

The Match of the Century

The isolated pawn, seen in Strategic Maps: isolated pawn, offers White various strategic possibilities. Among the most common plans, it can serve as a base for an attack on the King’s side, often through maneuvers with the Knight on e5 or even sacrifices to open lines and create initiative.

Another key idea is the d5 push, which, if played at the right moment, can free up piece activity and transform the static energy of the position into concrete dynamism.

In the match between Deep Blue and Kasparov, two Sicilian Alapins were played, with Kasparov always playing Black. Being one of the world’s foremost experts in the main Sicilian variations, it was natural for Deep Blue’s team to choose a solid line.

Deep Blue - Kasparov,G - 1996

Deep Blue - Kasparov,G - 1996

Position after 22...Qf6

The Match of the Century

The isolated pawn, as seen in Strategic Maps: isolated pawn, can be approached by Black in various ways. One of the most common strategies is to blockade it immediately, carefully controlling the d4-d5 push to limit its activity.

Alternatively, Black can aim for simplification, seeking to exchange pieces and head towards a favorable endgame, where the isolated pawn becomes a more evident weakness.

Black can choose to exchange a piece on c3, inducing bxc3. This eliminates White’s weakness on d4, but at the same time allows shifting attention to the backward pawn on c3, which could become a target, especially if White fails to create adequate dynamic counterplay.

Deep Blue - Kasparov,G - 1996

Deep Blue - Kasparov,G - 1996

Position after 16.Qxf3

AlphaZero Against Stockfish

AlphaZero defeated the strongest chess engine of the time, Stockfish 8, in a 100-game match in December 2017 (achieving 28 wins, 72 draws, and zero losses). This result marked a turning point, as AlphaZero did not use the traditional brute-force approach, but a neural network trained through self-learning.

“It’s a remarkable achievement…It approaches Type B, the semi-human approach to machines dreamed of by Claude Shannon and Alan Turing instead of brute force”

— Garry Kasparov

The idea of machines capable of playing chess without relying on pre-programmed rules has its roots in the work of the fathers of the digital age. Von Neumann, Wiener, Turing, and Shannon can be considered the precursors of modern artificial intelligence-based chess engines, like AlphaZero.

AlphaZero Against Stockfish

The Carlsbad structure, seen in Strategic Maps: Carlsbad structure, offers Black classic positional plans. As a general rule, Black seeks to respond to the Ne2, f3-e4 plan with c5, especially because f3 weakens e3 and d4.

However, in the match between Stockfish v8 and AlphaZero, the latter demonstrated that even in such an established position, creative plans can be found.

It demonstrated that it’s possible to evolve traditional strategy, introducing new creative elements even in such an established structure.

Stockfish - AlphaZero - 2018

Stockfish - AlphaZero - 2018

Position after 18.Rfe1

AlphaZero Against Stockfish

The Carlsbad structure, seen in Strategic Maps: Carlsbad structure, offers White classic positional plans like the minority attack or the central push to gain space.

In this second game, AlphaZero played a completely unexpected move, challenging the typical conventions of chess education and directly provoking the opponent.

It demonstrated that even from White’s side, these positions offer unexplored and creative possibilities, redefining traditional strategic plans.

AlphaZero - Stockfish - 2018

AlphaZero - Stockfish - 2018

Position after 9...Bg4

Beyond Calculation: The Human Factor

Example 1

Example 1

White to move

This is one of those positions where human intuition can still surpass machine calculation. If you analyze it with a chess engine, you’ll see it chooses to capture on b6, leading to a positional draw.

But can you find the winning plan for White?

Beyond Calculation: The Human Factor

Example 2

Example 2

White to move

This is also one of those positions where human intuition can still make a difference. Analysis engines are not unanimous in evaluating the correct plan: some identify the right path, while others hesitate.

The key here is finding the way to prevent Black from building a fortress and achieving a draw. Can you see how White can break through?

Beyond Calculation: The Human Factor

Movsesian,S - Kasparov,G - 2000

Movsesian,S - Kasparov,G - 2000

Position after 16.Bd3

This position recalls a theme already addressed in Chess and Creativity: the essence, where Black opts for a positional sacrifice, a choice difficult for an engine to evaluate.

In exchange for the quality, he obtains weakening of the opponent’s castled position, excellent outposts for light pieces that enhance the attack, and a more harmonious pawn structure.

Interestingly, while more advanced engines like Stockfish > 12 and Leela recognize the strength of this move, those from previous generations tend to ignore it, preferring other continuations. Can you evaluate whether Black’s compensation is sufficient?

Beyond Calculation: The Human Factor

Kasparov,G - Shirov,A - 1994

Kasparov,G - Shirov,A - 1994

Position after 16...Nc5

Finally, this position also recalls themes discussed in Chess and Creativity: the essence. White finds an extraordinary idea, unbalancing the position already on the 17th move.

His goal is to completely dominate the light squares, particularly the outpost on d5, which he occupies with the knight. To achieve this, he once again chooses a positional sacrifice, eliminating Black’s light-squared bishop and forcing his knight into an uncomfortable position.

More advanced engines recognize the idea, while those from previous generations overlook it.

Conclusions and Key Points

Artificial intelligence has redefined the game of chess, introducing new ideas and strategies previously unthinkable. However, human creativity remains an essential element, capable of exploring unknown territories and making decisions that escape algorithmic logic.

The Evolution of Artificial Intelligence

From the first brute-force engines to AlphaZero, the paradigm shift has led to greater “artificial intuition,” with more fluid and less forced strategies.

The Impact on Players

Engines have profoundly transformed preparation, offering new insights and redefining the very concept of creativity and innovation in chess. Some knowledge emerging from their analysis is completely new, bringing never-before-considered ideas and expanding the way we interpret the game.

Conclusions and Key Points

In-Depth Videos

Further Reading (1/5)

Deep Thinking

Deep Thinking

Author: Garry Kasparov

Publisher: PublicAffairs

Written by Garry Kasparov, this book examines his historic encounters against IBM’s supercomputer, Deep Blue. Kasparov reflects on his experiences during the crucial games, exploring the reasons for his defeat and the implications of human-computer interaction.

Further Reading (2/5)

L'Eterna sfida

The Eternal Challenge

Authors: Michele Godena, Bruno Codenotti

Publisher: Hoepli

M. Godena and B. Codenotti guide us on a unique journey through history, science, and technology, showing how chess programming and the computer revolution have walked hand in hand. The book explores figures like Turing, Shannon, and McCarthy, who laid the foundations for the automation of chess thinking and traces the evolution of chess software from the 1960s to AlphaZero.

Further Reading (3/5)

Gli scacchi tra fosforo e silicio

Chess Between Phosphorus and Silicon

Authors: Enrico Pepino, Nicola Vozza

Publisher: Le Due Torri

Published in 2010, this book analyzes the relationship between human intuition and artificial calculation, examining played games and comparing them with evaluations proposed by engines. The authors highlight how humans make decisions, emphasizing the differences from the cold and objective analysis of the era’s software, based on brute-force calculation. A fascinating journey between creativity and computational precision.

Further Reading (4/5)

Game Changer

Game Changer

Authors: Matthew Sadler, Natasha Regan

Publisher: New In Chess

Game Changer tells how AlphaZero redefined chess principles through a creative and unconventional approach. The book offers a collection of games, analyses, and reflections on AlphaZero’s impact on the modern game. It’s a guide for those who want to explore the combination of artificial intelligence and creativity, showing how AlphaZero’s ideas have inspired new strategies in positional and tactical play.

Further Reading (5/5)

The Silicon Road to Chess Improvement

The Silicon Road to Chess Improvement

Author: Matthew Sadler

Publisher: New In Chess

The book offers tools to expand strategic repertoire and deepen middlegame understanding, providing clear models for typical positions. It helps use analysis engines more effectively, explaining suggestions in accessible terms. Additionally, it offers insights into high-level play, making strategies and tactics adopted by grandmasters comprehensible.