Nowadays, a lot of people are interested in Artificial Intelligence (AI) because it is the technology of the future. A lot of people consider it a singular detached field that can be studied in isolation. But AI is the culmination of multiple disciplines like Computer science, Biology, Psychology, Linguistics, Mathematics and Engineering. Owing to its multifaceted nature AI has attracted attention from all walks of life. AI has developed further and branched out to Machine Learning (ML) and Deep Learning (DL) as well. However, these terms are used interchangeably by a lot of people not realizing what they mean.
To keep it simple, let us imagine the Earth which is a part of our solar system. Our solar system is a part of the Milky Way Galaxy. Similarly, if we consider Artificial Intelligence to be the galaxy and solar system to be Machine Learning. It naturally implies that the earth will be synonymous with Deep Learning. Deep learning is a subdivision of machine learning, and machine learning is a subdivision of artificial intelligence, which is an umbrella term for any computer program that can do something smart. In other words, all machine learning is AI, but not all AI is machine learning, and so forth.
So, the term Artificial Intelligence was coined by John McCarthy in 1956. Often considered to be one of the godfathers of AI, he described it as ‘the science and engineering of making intelligent machines’. Artificial Intelligence is generally classified into two branches, namely; applied and general. For example, applied AI refers to systems designed to intelligently trade stocks and shares, or manoeuvre an autonomous vehicle etc. Whereas general AI relates to systems/devices that can, in theory, handle any task. This is a developing branch of AI yet the most exciting one.
Next, is Machine learning. It is a subdivision of AI, and Arthur Samuel coined the term. He described it as ‘the ability(of a computer) to learn without being explicitly programmed’. Machine learning implies that it can modify itself when exposed to more data without requiring any human experts manually making changes. A great example of machine learning is how computer vision has improved over time. Let’s say human experts tag birds and cats in thousands or even millions of pictures. After that, this algorithm tries to build a model, distinguishing photographs of birds from cats or both of them. Once the level of accuracy is high enough, the machine can differentiate between birds and cats which means it has learned what they look like in pictures.
After that came Deep Learning which is a subdivision of machine learning. ‘Deep’ here is a technical term referring to the number of layers in a neural network imitating the human brain. For example, each layer in a neural network picks out a specific feature to learn, such as curves/edges in image recognition. These layers help in giving results with higher accuracy even though it requires more hardware and training time. It works best on machine perception tasks that involve unstructured data like piles of text or pixels.
These converging technologies together are accelerating development in this arena, which in turn will profoundly impact our society in the years to come.