There are a lot of new terms buzzing around the topic of AI. Each one may take on a specific meaning with subtle differences or describe the research or specific type of AI, but nevertheless are covered under the umbrella term of AI or Artificial Intelligence. Not only are there different levels of awareness or abilities:
- Type I: Reactive Machines- As the name states, they are “reactive” only. They have no ability to form memories or make informed decisions.
- Type II: Limited Memory- These intelligent machines are able to form “limited memories”. Think of an automated vehicle needing to observe and remember the behavior of vehicles around it.
- Type III: Theory of Mind- These machines will be able to form understandings of the world around them, such as other machines, objects, animals and humans.
- Type IV: Self-Aware- In order for this stage of machine learning to come to fruition, humans would actually need to understand consciousness in sentient beings. This is obviously a long way down the road since we are only beginning to really study the functions of the brain in-depth.
Ok, we now understand the different types of AI and we can safely say we are getting more comfortable with Types I & II. But, Types III & IV are still in the works and will be for quite some time. As the industry grows, there is no doubt that new categories and sub-categories will emerge. So, let’s explore a few terms used to describe AI that you may come across while you’re surfing the neural network innovation ocean (I am sure I have missed quite a few, though!):
- Thinking Computer/Computational Thinking — The thought processes (comprising of 4 processes: decomposition, pattern recognition, pattern generalization and abstraction, and algorithm design) involved in formulating a problem and figuring out the solution.
- Neural Network — An interconnected group of nodes, similar to the connections of neurons in the brain.
- Machine Learning — Uses statistics to teach a program how to “learn” from data, without having the solutions explicitly programmed into the system.
- Deep Learning / Deep Structured Learning /…