by Albert Major

Introduction to Artificial Intelligence

Key Concepts

Syntax

• Definition: Rules for manipulating and combining symbols
• Key characteristics:
  • Deals with form and structure rather than meaning
  • Involves formal, structured systems
  • Goes beyond just language rules
  • In computing context: how symbols and data are processed according to defined rules
  • Represents the "mechanical" aspect of processing
• Example in computing: How a program processes ones and zeros without understanding their meaning

Semantics

• Definition: The meaning and understanding behind symbols
• Key aspects:
  • Goes beyond mere rule-following
  • Involves subjective experience
  • Includes feelings and understanding about concepts
  • Cannot be reduced to pure symbol manipulation according to Searle
• Example: Understanding what a "cat" truly is, beyond just processing the symbol "cat"

Connectionism

• Definition: An alternative approach to traditional symbolic AI
• Key characteristics:
  • Based on neural networks
  • Processing is parallel rather than serial
  • Uses sub-symbolic processing
  • Meaning is distributed across the network
• Historical context:
  • Initially a marginal approach in AI
  • Became dominant with modern deep learning
  • Represents a contrast to symbolic AI
  • Key figures like Geoffrey Hinton were initially dismissed but later vindicated

Behaviorism

• Definition: Psychological theory focusing solely on observable behavior
• Key principles:
  • Only observable behavior can be studied scientifically
  • Internal mental states are considered unscientific
  • Still influences modern AI evaluation
  • Connected to Turing Test methodology
• Modern relevance:
  • Influences how we evaluate AI systems
  • Related to metrics of AI performance
  • Sometimes seen as outdated but principles persist