Artificial Intelligence

There are three major approaches to Artificial Intelligence

• Symbolic AI

• Statistical AI

• Subsymbolic AI

• AI as Rational Agents

  • Propositional Logic.
  • First-Order Predicate Logic
  • Knowledge Representation
  • ALC
  • Planning

• Rational Non-Deterministic Agents

Alternative AI Definition

• Artificial Intelligence (AI) studies how we can make the computer do things that humans can still do better at the moment.
• Once AI solves a subfield it is called computer science.
• Doing things better than humans
  • Systems that
    • think like humans → Cognitive Science, how does the brain work?
    • think rationally → Logic, formalization of knowledge, inference, planning, constraint…, what is the right way of thinking?
    • act like humans → Turing Test
    • act rationally → Contains Logic, making good choices in practice (doing the right thing), Rational Behavior, Rational Agent

Possible components of AI:

• the ability to learn
• inference
• perception
• language understanding
  • related: Natural Language Processing
• emotion

Automated Reasoning, Automated Theorem Proving Automatically verify / proof correctness of programms. For example Intel verifies all chips after “Pentium 5 Disaster”.

The AlphaGo Approach: Symbolic AI → Search Space → Use NN (Subsymbolic AI) to search through this search space

Producer Tasks → Symbolic AI Consumer Tasks → Subsymbolic AI