In this article we are going to go over the key differences between these topics and clear up a few misconceptions surrounding them. We have heard a lot of talk about these subjects and as can be expected of buzzwords they can be a little confusing or even misleading at times.
Otherwise referred to as AI, the original term was invented by John McCarthy circa 1956. This term is used to describe any machine or computer that can perform human like tasks. Chatting, recognizing objects, etc. Common examples include chat-bots and virtual assistants such as Amazons Alexa, Siri from Apple or Google Assistant on Android.
There are differences in the varying types of AI which we identify based on their intended use and abilities. A “general” artificial intelligence is any AI that is intended to do many different types of human like tasks. A “narrow” AI is meant to specialize in just one, or only a few very specific tasks. The common factor here being that these are human-like tasks.
Machine learning is not AI itself but is a process in which AI is programmed. What is meant by this term is that a program is written in such a way as to learn on its own when presented with enough examples of how it should behave. In this way an often incredibly complex program can be created without needing to manually create massive amounts of complex code. Instead the rules are determined initially, and the program learns how to do its job through the process of analysis. Machine learning is dependent on feeding new programs huge volumes of data to learn from. Data is essential in order to make choices and assumptions based on past examples.
If you have ever bumped into a captcha that asks you to identify a specific object in a group of photos, or maybe identify some difficult to read words in a handful of images then you have probably helped teach a bot to think for itself!
Deep learning is a specific form of machine learning. There are others such as decision tree learning, inductive logic, reinforcement learning, etc. but for today we are just going to talk about this one.
Deep learning is a bit of a buzz word lately, and for good reason. It is particularly interesting because it works similar to the way our own brains work. Using the model of a neural network, a program is written with algorithms that each operate like an individual neuron. Each one has learned a hyper-specific part of a larger body of processes and is intended to be accessed as a component of a larger collection. In this network of similarly small parts, individually each process may be very simple but as a whole the entire program can be highly intelligent.