Artificial Intelligence is a term we find nowadays almost everywhere, websites, apps, books, articles, news, everywhere… But what is it? What is AI?
First things first, to begin with, artificial intelligence is a so-called “buzz-word” and here is why: It sells easier than “machine learning”. If you want to sell your application, or promote your CV, use “AI” instead of “Machine learning”. People will be more intrigued, because even if they don’t actually know what it means, it sounds hype.
What is Machine Learning?
In the beginning of the era of computers, a human had to sit down and tell it bit by bit what to do. For example, if you wanted it to solve an equation, you had to program it step by step to do so. But machine learning introduces the concept of computers that can learn by themselves (e.g. to solve the equation).
Machine learning refers to the computer’s ability to learn certain “skills”, with limited human intervention. But to do so, you need to present it with a set of inputs that it can learn from. It works kind of like our traditional education system. In class, you learn numbers, then you learn simple additions and multiplications, like 2 + 2 = 4 or 2 x 2 = 4. This is your set of inputs. Then you go home and your homework is to solve ( 2 + 2 ) x ( 2 x 2 ) = ?. By having your initial knowledge, you can deduct the result of this computation, even if you have never seen it before in this form. This is a simplified explanation on what machine learning does. The most common application of machine learning are neural networks. Where this “knowledge” is put to work.
What is a Neural Network?
A Neural Network, also called Artificial Neural Network (ANN) is an algorithm inspired by the human brain. It contains artificial neurons for processing information. The information propagates through the network from the input node to the output, and may or may not backpropagate, similar to a feedback loop, kind of like learning from past mistakes and getting better next time. A common type of an artificial neural network is the convolutional neural network.
What is a Convolutional Neural Network (CNN)?
A CNN is a neural network specialized in visual imagery. The term convolution comes from certain mathematical operations applied to an image in the form of “sliding windows”. These operations convolve on the image and perform tasks that are programmed in the architecture of the CNN.
With these definitions at hand, (and programming knowledge, of course) one has the power to teach a computer to recognize unknown objects, to “predict the future” by analyzing patterns, to decode series of emotions, gestures or speech and language. This is the so-called AI, this is artificial intelligence. With it becoming ever more popular, it is important to get the basic understanding of how it works, before you can analyze its risks.