GAMES
AND AI
By Khushboo Hundlani
Fintech 3rd sem
INTRODUCTION TO GAME
PLAYING AND AI
Game playing is a key domain in AI.
Games provide structured environments
with clear rules, making them ideal for
training algorithms.
AI in games has applications in game
development, education, and military
training.
CHARACTERISTICS OF AI
IN GAMES
- AI learns, adapts, and makes strategic
decisions.
- Helps develop decision-making
systems for real-world scenarios.
- Common technique: Minimax Search
Procedure (used in games like Chess
and Tic-Tac-Toe).
MINIMAX ALGORITHM
OVERVIEW
Key features of the Minimax Algorithm:
Depth-first, depth-limited search.
MOVEGEN: Generates possible moves.
STATICEVALUATION: Rates the quality of
a game state.
Player1 maximizes, Player2 minimizes,
recursively calculating optimal moves.
EXAMPLE OF MINIMAX IN
ACTION
1. Player1 starts the game; values at
nodes represent state quality.
2. Alternating levels maximize or
minimize values based on player's turn.
3. Final calculation suggests Player1's
optimal move to maximize the chance
of winning.
ADVANTAGES OF AI IN
GAMES
Drives advancements in AI algorithms.
Educational and training benefits.
Active area of research for decision-
making and problem-solving.
Real-world applications include robotics
and autonomous systems.
DISADVANTAGES OF AI IN
GAMES
Limited scope: Techniques may not
apply directly to other domains.
High computational cost, especially for
complex games like Chess or Go.
OPTIMAL DECISION
MAKING IN GAMES
Games require decision-making under
constraints.
Example: Chess involves an enormous
search tree, emphasizing efficiency.
Decisions modeled as a search problem
with states, actions, and utility values.
ALPHA-BETA PRUNING:
OPTIMIZATION OF MINIMAX
Alpha-Beta pruning reduces the number
of nodes examined.
Alpha: Best choice for maximizer; Beta:
Best choice for minimizer.
Pruning conditions: Skip branches where
Alpha >= Beta.
ALPHA-BETA PRUNING
EXAMPLE
Start with initial Alpha and Beta values.
Compare and update values at each
level.
Prune branches that don't impact the
final decision.
Improves efficiency without
compromising decision accuracy.
ROLE OF STOCHASTIC
GAMES IN AI
Stochastic games involve randomness
and multi-agent interactions:
Transition probabilities influence state
changes.
Common in environments requiring
decision-making under uncertainty.
APPLICATIONS OF
STOCHASTIC GAMES
- Autonomous systems: Self-driving cars
and robotics.
- Financial markets: Optimizing
strategies under market volatility.
- Healthcare: Treatment planning and
resource allocation.
- Security systems: Predicting and
mitigating cyberattacks.
EXAMPLE: STOCHASTIC
GAME IN BACKGAMMON
1. Combines skill and chance with
random dice rolls.
2. Legal moves depend on dice
outcomes, modeled as chance nodes.
3. Expected values guide decision-
making.
GAME TREE AND
DECISION MAKING
- Nodes represent game states; edges
represent actions.
- Utility values determine optimal
strategies.
- Example: MAX and MIN players
alternate moves to maximize/minimize
payoffs.
MOVE ORDERING IN
ALPHA-BETA PRUNING
- Ideal ordering prioritizes best moves
early.
- Rules for good ordering:
1. Explore shallow nodes first.
2. Leverage domain knowledge (e.g.,
Chess heuristics).
CONCLUSION
- Games are a benchmark for AI
research and development.
- Techniques like Minimax and Alpha-
Beta pruning enhance decision-making.
- Applications span entertainment,
education, and real-world problem-
solving.
