The difference between Artificial Intelligence or AI and Machine Learning is more fundamental than what we see in practice. Lots of people use the terms, AI and machine learning interchangeably but in fact, they are not quite the same. Artificial Intelligence is basically the umbrella term and within it, we have Machine Learning which is a subset of AI. Machine Learning is basically the leading edge of artificial Intelligence. We, at Oodles, as a well-established AI Development Company, share insights into the origins of AI and ML as being evolving since the 1950s.
The thing with Machine Learning is that it is probably ahead of its time and what we had is we had traditional artificial Intelligence tools. These tools were basically expert systems that involved telling computers exactly the rules on how to analyze data, what data to use, and what results to spit out. So, we had expert systems that worked sometimes quite well but we have this typical problem of computer says no just did not work or it did not know the answer, so for me, a great example is Natural Language or language translations so if we try to design a rule-based Artificial Intelligence program to translate from English to other languages, this does not work and we have seen this in the past that these things did not really work that well because there are so many exceptions to our human language and to program all these exceptions into this algorithm is almost impossible.
Image source: https://towardsdatascience.com/artificial-intelligence-vs-machine-learning-vs-deep-learning-2210ba8cc4ac?gi=99d03076340c
Going back to 1950s, Someone then thought actually instead of telling the computer all the rules why don't we give the computer lots of data so the computer can make up the rules by itself and this is what we are referring to as machine learning development where the machine learns from data and this is a bit like how we learn by ourselves.
This basically simulates the brain copies the process that we as humans use to learn and be intelligent so in our head, we have a brain and this has trillions of neurons and these neurons are all connected and when we learn something so you might learn how to grab something in a Sur or how to speak a language and as a child it takes quite a long time to do this and we go through lots of trial and error so we learn by experience and how do I grab this toy for example.
Your neurons make connections you say sending these signals to these muscles really worked in the same way when you pick up a language your parents and the teachers will correct you and over time you will learn how to speak the language. You can not really learn this by rules you learn this by experience. The challenge is that learning a language or grabbing a toy or cycling on a bike are things we can not actually explain. They are what we call it as Tacit Knowledge we have explicit knowledge that we can write down on a piece of paper I can write down this is how you operate a camera give this to you. You can read this and then operate the camera. I can not do this with this is how you swim or this is how you cycle this is how you speak English because this is something we have learned through experience and this is exactly what machines are able now to do. We give them data and they learn from this data so initially, we had tools that could recognize characters, and we all right handwritten characters differently. But now we can give a machine a million or billion version of how someone writes an O and an A and a T or whatever and then the machine learning algorithm will say Okay I know to build my own system and identify how probable this that this is an O or T. So this is again not a brand new since the 1960s.
In 1965, I think the US Postal Service implemented its first-hand writing scanner in their Detroit post office that is able to read someone's address on a letter handwritten letter and this helped them to improve things. What we now have is we have the machine learning capabilities, Why do we have them? today two things have changed. We now have more data because we are now living in the big data world where we have lots of sensors. Everything is digital so we have huge volumes of data and we have the processing power so our chips are getting better and we have things like cloud computing that gives every device access to huge computing.
The ability to store huge volumes of data and analyze them and this is now made machine learning possible so instead of saying this is how you translate from English to any other language you simply give the machines billions of words and text translated from other languages into English and then the machines will write their own algorithms to be able to do this and this is basically the leading edge of Artificial Intelligence. It now enables machines to learn to walk to learn to write to we now have tools like Natural Language and voice recognition tools like Alexa that can pick up our language whether we are speaking with an American accent or any other accent and this is all made possible by machine learning and improving.
Hopefully, this has given you a better understanding of what is the difference between Artificial Intelligence and machine learning which is the rule-based overarching concept of AI and the more specific leading-edge application of machine learning. The ability of machines to form data a bit as we learn from experience.