Artificial intelligence and machine learning (machine learning) are two concepts that appear every time they are heard more and sometimes they can be confused. With this post we want to clarify the difference between the two while defining them.
Artificial intelligence (AI) is a very broad concept that refers to any tool, algorithm or technology that allows a machine to have typical functionalities of the human brain. We could say that goals of artificial intelligence they would include reasoning, planning, learning, language processing, perception of objects and the ability to move them. These would be short-term goals, and there are already many advances in several of these fields.
Within artificial intelligence we would have more ambitious objectives such as providing machines with the ability to solve any type of problem, which would be a longer-term objective and which would be within what is called strong artificial intelligence. Whereas when we focus on just one of the human capabilities (such as vision or language), we call it weak artificial intelligence.
Machine learning or machine learning (ML) it is one of the tools to achieve that artificial intelligence. As its name makes us intuit the algorithms of machine learning they are based on learning about a set of examples. For a machine the examples are given with a lot of data. They can easily learn to classify objects, recognize pictures, or make predictions. Models of ML tools are often called predictive models. There are many proposed ML algorithms that work very well, one of the most widely used being neural networks. Here You can see a list of the most typical ML algorithms.
Artificial intelligence without machine learning
To better see the difference between AI and ML, we propose some examples of artificial intelligence without using machine learning.
1) A machine with sensors that detects movement and has certain basic instructions to interpret and react could be said to be artificial intelligence without having used any ML technique.
2) A computer where we can dump the knowledge of many experts from different areas of a certain field. This could allow you to answer all kinds of questions in that field, better than just one of the experts. This would fall within the objectives of artificial intelligence, in the sense that it emulates human reasoning, but the machine itself has not done a learning process, so it has not used ML. This type of computer system is called expert systems, and although they can be useful in some specific fields, artificial intelligence aspires to much more.
The deep learning revolution
Artificial intelligence is now often associated with ML because since we are able to collect and store large amounts of data, ML techniques have been able to exploit their full potential. This is because with these large databases we can train and make machines learn very well with these algorithms, since we give them many examples. One of the tools that has given the best results is what is known as deep learning, which are algorithms based on neural networks adapted to solve complicated problems with large amounts of inputs (for example in images that contain a high number of pixels).
In Figure 1 you can see what the hierarchical organization of these three subgroups that we have talked about would look like: artificial intelligence, machine learning and deep learning.
Figure 1. Hierarchical diagram of artificial intelligence, machine learning or machine learning and deep learning.