My description of machine learning using only the ten hundred most used words (via the up-goer five text editor):
This is the study and work of making computers learn. It has three important parts. First, it needs number pictures of how things could be. Second, it needs a way to see how good these pictures are. Finally, it needs a way to find the best picture from the group of possible number pictures. Once the best picture is found, it is used to guess what new things will be like or should be like. This helps us do many cool things like suggesting stuff you might enjoy using a picture of what you like; help you write using a picture of good writing; stop bad things before they happen by seeing that they are like our picture of what is bad, and lots more.