Our data is a very modern raw material
Nowadays, the practical approach has taken the lead in comparison to the theoretical one. Relatively often (especially when it comes to Deep Learning) there are many things that "work", but it is not entirely clear why... The only proof on which they are based is the fact that several or hundreds of other options have been checked and this one works best :) The real blossom of so called artificial intelligence (or actually machine learning, or in most cases - Deep Learning) has taken place in the last decade.
There are three key elements for activating machine learning: data, computing power (infrastructure) and algorithms. Of course, there is also a team at the center that combines all these elements. I will pay particular attention to the fact that computing power and algorithms are easily accessible today (provided by large players such as Google, Microsoft or Amazon). That's why we have so many applications that we can download "for free", and in fact their purpose is to collect user data (please think about how many "free" applications you use). I encourage you to listen to the second episode of the Business Thoughts podcast. Machine learning - a way to teach machines without giving rules directly. In a word, we can say that machine learning is optimization. However, let's take a step back first.
The programmer who solves the task must know the whole algorithm step by step - all possible cases. If there is no case, then frustration appears. By the way, this is one of the biggest difficulties in communication between programmers and business people, because they operate at different levels of accuracy. But a progamist is also partly a player. I'm not talking about mindlessly patting games. There are games that make you think more.