Artificial Intelligence Introductory Lecture

Introduction

• Welcome to the course
• Remarks on course demographics
• 10% turnover expected in the first 24 hours
• Agenda for today's lecture:
  • Definitions of AI
  • History of AI
  • Course covenants (e.g., no laptops)

What is AI?

• Involves thinking, perception, and action
• Models are important in understanding AI
  • MIT's approach: building models using various methods (differential equations, probabilities, simulations)
• Definitions:
  • AI: Representations supporting models for thinking, perception, and action
  • Representation: How information/data is structured

Importance of Representation

• Example: Gyroscopes and right-hand screw rule
• Representation helps understanding and prevents mistakes
  • Use correct visualization for problem-solving
• Example: Farmer, Fox, Goose, and Grain problem
  • Correct representation (state diagrams) exposes constraints and simplifies problem-solving

AI Components

• Methods/Algorithms/Procedures: Supported by constraints exposed by model representations
  • Example: Generate and Test
    • Generates possible solutions, tests them, discards failures
    • Vocabulary (naming) gives us power over the concepts (Rumpelstiltskin principle)
  • Simple ideas can be powerful

Purpose of AI

• Engineering perspective: Build smarter programs
• Scientific perspective: Understand computational nature of intelligence
• Dual focus: Practical application (smarter programs) and theoretical understanding (human intelligence differences)

History of AI

• Lady Lovelace (1842): First programmer, predictive statement about computers
• Alan Turing (1950): Turing test
• Marvin Minsky (1960): Steps toward Artificial Intelligence paper
• Jim Slagle: Symbolic integration program
• ELIZA: Early AI conversational program
• Geometric analogy problems: Symbolic reasoning challenge
• Rule-based expert systems: Examples like MYCIN for medical diagnosis, Delta Airline airplane parking
• Deep Blue: Chess-playing computer
• Current: Integration of perception, action, and thinking

Current Trends and Future of AI

• Shift towards understanding loops connecting thinking, perception, and action
• Example programs that simulate scenarios and use visual memory
• Evolutionary uniqueness of human intelligence:
  • Speculative cause: Ability to combine concepts via language

Course Structure

• Lectures: Introduce material, big picture
• Recitations: Twist and expand on material
• Mega recitations: Past quiz problems, skill-building
• Tutorials: Homework help

Attendance and Performance

• Correlation between lecture attendance and course grades
• Regression line indicating higher scores with higher attendance
• Grading mechanism: Scores converted to a 5-point scale, max of quiz and final part score

Administrative Notes

• Need to fill out tutorial schedule form
• Possible Python review session on Friday
• No regular recitations this week