Artificial Intelligence Principles, History, and Problem Solving

Flashcards covering AI definitions, capabilities, functionality, history, problem formulation, production systems, state space search, classic AI benchmark problems, and uninformed search algorithms (BFS and DFS).

What is Artificial Intelligence (AI) as defined in computer science?

Artificial Intelligence is the branch of computer science that aims to create machines capable of intelligent behavior, mimicking human cognitive functions like learning, reasoning, visual perception, speech recognition, decision-making, and problem-solving.

How is AI categorized under Type-1 (capabilities) versus Type-2 (functionality)?

Type-1 classifies AI by capabilities into Narrow (Weak) AI, General (Strong) AI, and Super AI. Type-2 classifies AI by functionality into Reactive Machines, Limited Memory, Theory of Mind, and Self-Awareness.

What defines Narrow AI (Weak AI) and what are some prominent examples?

Narrow AI is designed and trained to perform a single dedicated task within pre-defined limitations and can fail if pushed beyond its limits. Examples include Apple Siri, IBM Watson, Google Translate, Deep Blue, and recommendation systems.

How do General AI and Super AI differ in their concepts?

General AI (Strong AI) refers to systems that can think, learn, and perform any intellectual task efficiently like a human across any situation (still under research). Super AI is a hypothetical level where machines surpass human intelligence across all cognitive properties, including reasoning, planning, emotion, and creativity.

What is a Reactive Machine in AI functionality, and what are two key examples?

A Reactive Machine is the simplest AI type that does not store memories or past experiences for future decisions; it only evaluates the current scenario to choose the best action. Examples include IBM's Deep Blue and Google's AlphaGo.

How do Limited Memory AI systems operate?

Limited Memory systems store past experiences or observational data for a short duration to make decisions, such as self-driving cars tracking the recent speed and distance of nearby vehicles, speed limits, and road layouts.

What are Theory of Mind AI and Self-Awareness AI?

Theory of Mind AI aims to understand human emotions, beliefs, and social interactions (currently under active research). Self-Awareness AI is a hypothetical future AI possessing its own consciousness, sentiments, and self-awareness superior to the human mind.

What key foundation papers in AI history were published in $1943$ and $1950$?

In $1943$, Warren McCulloch and Walter Pitts published "A Logical Calculus of Ideas Immanent in Nervous Activity," creating the first mathematical model for neural networks. In $1950$, Alan Turing published "Computing Machinery and Intelligence," proposing the Turing Test.

When, where, and by whom was the term "Artificial Intelligence" officially coined?

The term was coined in $1956$ at the Dartmouth Conference organized by John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon.

What major AI milestones were accomplished in $1997$, $2011$, and $2016$?

In $1997$, IBM's Deep Blue defeated world chess champion Garry Kasparov. In $2011$, IBM's Watson won "Jeopardy!" against human champions. In $2016$, DeepMind's AlphaGo defeated world champion Go player Lee Sedol.

What are the five core components of a formulated AI problem?

The five components are Initial State, Actions (Operators), Transition Model, Goal Test, and Path Cost.

How are problems classified based on the severity of errors or failures?

Problems are categorized as Ignorable Problems (minimal/no impact on performance), Recoverable Problems (errors can be undone or recovered via mechanisms like checkpoints), and Irrecoverable Problems (critical failures leading to permanent loss or dataset corruption).

What is a State Space Search in AI?

It is a problem-solving technique that systematically explores all conceivable configurations (states) and permissible moves (actions) from an initial state to find a path to a goal state.

What are the primary components of a Production System in AI?

A Production System consists of a Global Database (central data structure), a Set of Production Rules ("if-then" rules operating on preconditions), and a Control System (selects rules to apply and resolves conflicts).

What four features characterize a Production System?

The four features are Simplicity (readable IF-THEN rules), Modularity (discrete knowledge units that can be added or deleted without side effects), Modifiability (rules can be easily adapted), and Knowledge-Intensive (knowledge stored directly in human-readable form).

What two criteria must any effective Control Strategy satisfy?

A Control Strategy must cause Motion (ensuring state changes that progress toward a solution) and be Systematic (preventing repeated useless cycles or loops).

What is the difference between Monotonic and Non-monotonic Production Systems?

In a Monotonic Production System, applying a rule does not prevent the later application of other rules that could have been applied at the same time. In a Non-monotonic Production System, applying a rule can invalidate or prevent the subsequent application of other previously valid rules.

How are the state space, initial state, and goal state defined in the Water Jug Problem?

States are represented as a tuple $(x, y)$, where $x$ is the gallons in the 4-gallon jug and $y$ is the gallons in the 3-gallon jug. The initial state is $(0, 0)$ and the goal state is $(2, y)$.

What are the goal state and optimal path cost in the Missionaries and Cannibals Problem?

The goal state requires all 3 missionaries, 3 cannibals, and the boat to be safely transferred to the right bank without cannibals outnumbering missionaries on either bank. The optimal solution path cost is $11$ crossings.

How are moves represented in solving the 8-Puzzle problem?

Instead of moving numbered tiles into blank spaces, a move is visualized as swapping the blank space with an adjacent tile in one of four orthogonal directions: UP, DOWN, LEFT, or RIGHT.

What is the heuristic function formula defined for Tic-Tac-Toe?

$\text{Heuristic function } H = \text{X's probability win} - \text{O's probability win}$, where $H = 1$ represents an X win, $H = -1$ represents an O win, and $H = 0$ represents a draw/tie.

How can a final solution state be represented array-wise in the 8-Queens Problem?

A valid state can be represented as a 1D array of column positions for each row, such as $[1, 5, 8, 6, 3, 7, 2, 4]$, where each element specifies the column location of the queen in rows 1 through 8.

What data structure does Breadth-First Search (BFS) use and what are its key properties?

BFS uses a First-In-First-Out (FIFO) queue data structure. It is complete (guaranteed to find a solution if one exists) and optimal for unweighted graphs (finds the minimal path length).

What data structure does Depth-First Search (DFS) use and what are its primary trade-offs?

DFS uses a Last-In-First-Out (LIFO) stack data structure. It requires significantly less memory (storing only nodes on the current path from root), but it is not guaranteed to find an optimal solution and can get trapped in infinite loops.