Kannan Chandrasegaran, a software developer, has written a very good non-technical guide to machine learning on his Medium blog.
He starts with something that most of us techs and non-techs have heard of recently, Artificial Intelligence, and he explains that nowadays lay people label anything that a computer does and seems somewhat smart with AI, which should not be the case.
Kannan uses an example of computer games, a problem that can be solved using rules, to introduce machine learning. However, ML is the opposite of that because in ML it is more likely that we do not hard code rules for a specific problem to be solved.
Rather, we use an algorithm that finds its best way to solving a problem. And one of the best examples that I can think of is computer vision.
Let's assume we have 1,000 images of cats and non-cats. We know that each image is either a cat or a non-cat. We want our ML program to learn that too. So, we feed the program these images and we tell it which are cats and which are non-cats.
If everything works well, our program will learn (find ways to make associations between the features in the image and the label 'cat' or 'non-cat'). After the learning is done, we can test our algorithm on new images and see how it performs. And this is an example of Supervised Learning, a very hot field of machine learning.
In Kannan's guide you learn about this type of learning in much more detail and with additional graphics to help you understand the concepts better. If you're interested, please read the rest by following the link below:
To stay in touch with me, follow 
Cristi Vlad Self-Experimenter and Author