What Is Machine Learning and Why You Should Care
A Non-Technical Introduction
Amazon can suggest interesting products, Netflix can propose new films, Siri can understand your speech, your phone camera can detect faces and ChatGPT can answer your questions.
These are only some examples of tools or services based on Machine Learning algorithms.
While media are focusing on chatbots, built on what we call Large Language models, ML algorithms are permeating our lives in many other areas, even if you don’t see them.
Even if ML is not interesting for your life or your work, you’ll interact with Machine Learning systems more and more often in the future. That’s for sure.
Machine Learning applications are here to stay, but why?
The Core idea
To answer the question, we first need to understand what an ML model does and how it‘s possible.
A Machine Learning model is a tool where you insert some data and obtain new data, which are answers to a given problem (also called predictions).
Let’s take the example of real-time face detection when taking photos. Your phone passes an image to an ML model, which will return the information needed to locate every face in the frame.
This is possible because that model was previously trained to recognize faces in photos. This means there was a set of photos where faces were already annotated, by hand, by people. That set was given to the model to learn the knowledge needed to complete the task.
Once the model is trained, we can give it a new photo and it will return the predicted face annotations. This process is called inference.
Now you should understand why data are the fuel of Machine Learning.
Data is the new oil
This is a famous phrase you have surely read somewhere else. Actually, I didn’t know Clive Humby coined it, in 2006. Eighteen years have passed and it is more actual than ever.
According to the latest estimates, 402.74 million terabytes of data are created each day.
So, to put the ingredients together:
- ML models allow the automation of many hard tasks
- data are the main ingredient needed to train ML models
- we live in an epoch where tons of data are produced every day, about everything
This explains why there has been a Machine Learning boom in the last years. But it also suggests that the possible use cases are infinite and probably there is one for your business too.
Demand forecasting, predictive maintenance and customer or employee churn prediction are only some examples.
Even if it isn’t your case
Surely I don’t want to convince everyone to become an ML engineer. In fact, the market has probably already started to saturate. Neither I want to persuade you that your business needs Machine Learning. Those are conclusions you should draw yourselves.
My point is that you’ll come across ML whether you like it or not, because many of the services you use will incorporate it. Maybe some tools you use to do your work will adopt ML algorithms in the future. Whatever the case is, it makes sense to have a basic grasp of how it functions.
It’s better to be prepared.
Thank you for reading. If you like seeing things through the lens of data, consider subscribing to not miss my next articles, it’s free. See you around.
