The difference between Artificial Intelligence, Machine Learning and Deep Learning

Source: Deep Learning on Medium

AI is undoubtedly the buzzword of the 21st century. But along with AI terms like machine learning and deep learning also gets associated. Even though these words are used synonymously they definitely don’t mean the same.


The term AI was coined by John McCarthy in 1956 and could be defined as giving machines the ability to reason like men. The term AI is a very broad category which includes the interaction of implicit as well as explicit variables to get the job done in an intelligent manner. AI can be classified into general and shallow. General AI can be exemplified in fiction by 3PO from star wars movies, and narrow AI which can perform specific tasks such as speech recognition, image recognition, object tracking etc.


On the contrary machine learning is a paradigm which aids AI by understanding patterns and insights from data, it basically doesn’t need external stimulus and can learn on its own. It avoids hardcodings from routines and subroutines and can understand the information it receives. Here the useful information can be extracted from the data in many ways like supervised learning or unsupervised learning or feature engineering. Examples include anomaly detection, spam filtering and so on.


Deep learning is a subset of machine learning where the inspiration is functioning of the human brain. This involves creating mathematical models of neurons and creating artificial neural networks (ANNs) and the number of layers of the neural network indicates the depth of network which in term leads to coining of ‘DEEP’ learning. This specific paradigm as perceived by many is not a new area. In the 80’s it was known as cybernetics but due to computational intensive requirements was lying dormant till 2012 when Alex Krizhevsky famously won the magnet competition (a competition on visual recognition) with an error of 15.3%, more than 10.8 percentage points ahead of the runner-up. With the advent of high performing GPU’s and CPU’s deep learning is blooming again. Typical examples of usage of deep learning including face recognition systems, fingerprint identification in smartphones and the list go on.

Note: this article was originally published here: