Definition of AI

• Cognitive Modeling: Thinking Humanly (study human thinking, mimic humans)
• Turing Test: Acting Humanly
• Laws of Thought: Thinking Rationally
  • Rule-based approach, rational input → rational output.
  • Difficulty: Hard to accurately model real-world in logical statements
• Rational Agent: Acting Rationally
  • Agent: Something that acts
  • Acts to achieve the best expected outcome
  • Pro: Rational behavior is more general than rational thought (you can have rational behavior without rational thought)
  • Pro: Rational is better than mimicking human because…
    • Human often act predictably irrational
• Reflex Agent:
  • Choose action based on current perception of each step, no future planning

Components for Search

1. Set of states
2. Initial state
3. Successor function (with actions & costs)
4. Goal states / goal test fn
5. (Optional) Cost for each action
6. (Optional) Heuristic function as a guide

Uninformed Search

Planning Agents: Decide based on hypothetical consequences

• Have a goal, know the outcome of each action
• Optimal Planning: Minimize a cost function when completing the goal
• Complete Planning: If there exists a solution, will we find it
• Replanning: Breaking down a problem into smaller problems

Uninformed: Brute force, no additional info about the goal