What is the difference between Artificial Intelligence and Machine Learning?
Artificial Intelligence is intelligence induced artificially. So when we speak about AI, we are referring to the broad set of cases where the intelligence of a computer is used to solve problems.
Over the years many ways were developed to achieve artificial intelligence, they can be broadly divided into two categories Symbolic and Non-Symbolic.
Symbolic AI
This way of approach is surprisingly deterministic, which kinda kills the basic motive of AI that a problem should not be explicitly programmed. This is an intelligent way of looking at the bigger patterns underlying within a set of programs and then programming the system based on the identified patterns.
The best example of the use of this approach in day-to-day life is the use of dictionaries.
If you were given a phrase
exasperating farrago of distortions misinterpretation and outright lies
you might not know the meaning of any of the words, but using a dictionary
exasperating = intensely irritating
farrago = a confused mixture
distortion = misleading
can give you a basic understanding of the phrase. This approach of using symbols as the basic entity is called Symbolic AI. In this specific example, a word is a symbol.
This was the first approach to have experimented in the field of AI because setup is minimalistic i.e, there are no layers of abstraction, where human readable and understandable symbols are used. Later this approach was discarded because this cannot be used for bigger problems such as translating a book, where there are many factors such as context, sentiment etc, and complex grammatical structures like idioms. In such a case an innate understanding of a language is required rather than having a dictionary.
Machine Learning
Machine Learning is a type of non-symbolic approach to AI, which often uses statistical techniques to build learning algorithms.
For example to solve the following problem
given a set of points in 2D plane construct a straight line such that the loss is minimum
Unlike, the symbolic approach we do not have or use a compiled dictionary with different sets of points and corresponding straight line with minimal loss, the machine learns through a learning algorithm based on the statistical definition of loss of data. The algorithm used in this example is called Linear Regression. It is the fundamental learning algorithm in Machine learning, based on which many other ML algorithms were built.