AI for Good

Chapter 2Artificial Intelligence: Its Application and Limitations

—Juan M. Lavista Ferres

In many ways, “artificial intelligence” is a marketing term that has evolved over time. Even back in the 1970s, some applications that had a few rules were considered artificial intelligence. Today, there are many debates about what qualifies as artificial intelligence. In practical terms, most of the time when we talk about artificial intelligence, we are referring to machine learning.

So, what is machine learning? It is a technique involving algorithms that transform data into rules. In conventional programming, humans use their intelligence and knowledge to create rules expressed in software code. In contrast, machine learning relies on data and success criteria, using this data to generate rules while optimizing that criteria.

Many say that data is the new oil. In reality, data has nothing to do with oil; data is the new code.

Machine learning methods are not new: they are older than computers. We didn't refer to them as machine learning in the past, but as statistical methods. This includes techniques such as linear regression, logistic regression, and linear discriminant analysis (LDA). Some of these methods, like linear regression, date back to the early 19th century.

As an example, imagine you want to predict the price of a house. A basic model might use just square footage. By collecting data on square footage and sale prices, you can plot them on a chart (see Figure 2.1). The ...