Approaches to AI: Combination of three sources

  • Posted on: 19 December 2017
  • By: Juho Vaiste

How to define and more importantly how to perceive the different approaches of AI remains as quite a ragbag.

Here's the definition of the different approaches of three relatively well-known sources. Yes, I have included Wikipedia here, Wikipedia articles on AI are relatively good and clear at least in giving the overview on the subject.

I still struggle myself to see the overall picture and structure clearly enough. The idea of this post is just to continue processing that further and perhaps building a summary of the approaches later on - from my structure-loving and systematic point of view.

Cambridge Handbook of AI: Approaches to AI

1. GOFAI (Good Old-Fashioned Artificial Intelligence)
2. Connectionism and neural networks
3. Dynamical systems and embedded cognition

Wikipedia, the overall article "Artificial intelligence"

1. Cybernetics and brain simulation

2. Symbolic AI
- Cognitive simulation
- Logic-based
- Anti-logic or scruffy
- Knowledge-based

3. Sub-symbolic
- Embodied intelligence
- Computational intelligence and soft computing

4. Statistical

5. Integrating the approaches
- Intelligent agent paradigm
- Agent architectures and cognitive architectures

Artificial Intelligence - A Modern Approach

The book is built on the idea and concept of an intelligent agent.

1. Acting humanly: The Turing Test approach
2. Thinking humanly: The cognitive modeling approach
3. Thinking rationally: The “laws of thought” approach
4. Acting rationally: The rational agent approach