As of 2019, we are all familiar with the term artificial intelligence. However, those two words have taken on an incredibly vague meaning. And there is a good reason too. Check any two different dictionaries for the term and there is a strong probability that their definitions will vary.
One reason for this variation is the wide variety of uses that fall under the AI definition. Children's toys such as the old school Teddy Ruxpin is considered to use AI the same way StormGeo is using AI to make the most accurate weather predictions in history.
However, these two examples diverge in two different directions when observed closely. The adorable Teddy Ruxpin's programming is complete. It can only fetch and display data. The only way it can improve on itself is with an entirely new model. Contrarily, the weather predicting software by StormGeo uses a subset of AI known as machine learning which continually improves upon itself to make better and better predictions and decisions.
So What Is Machine Learning?
Machine learning is a system that improves upon itself by learning from examples, using algorithms to create patterns from the examples given to it. The more narrowly defined the pattern becomes, the better the AI will become at solving its objective.
Essentially, machine learning helps programs find optimal separations by maximizing correct classifications while minimizing errors.
Programs using machine learning use features in order to become better classifiers. Features are values that usefully characterize the things we wish to classify.
Take Netflix for example. The AI software uses a type of machine learning in order to predict which show or movie you would like to watch next. They even show you the numerical output of their system by giving the user a "match" percentage.
They will be wrong sometimes, which is why the match is never 100%. However, the more you watch (data input), the better this prediction algorithm will become.
Another example is your Facebook timeline. Facebook uses machine learning in order to make decisions about who's posts reach the top of your news-feed. They also use machine learning in order to predict what kind of advertisements you would like to see. If you're curious about what they came up with, you can check it out for yourself at https://www.facebook.com/ads/preferences.
"When I see a bird that walks like a duck and swims like a duck and quacks like a duck, I call that bird a duck."
-James Whitcomb Riley
The quote above truly simplifies what machine learning is all about. However, systems today are much more sophisticated and able to take features and analyze them by the thousands at speeds not possible just ten years ago.
The faster our processing speeds get and the smarter our machines become, we will find more and more useful applications for AI and this type of data as our society becomes more reliant and demanding of getting more actionable and useful outcomes for this type of technology.