Math is the foundation of AI

- If you want to understand and develop AI, you need to know mathematics, says doctoral student Taariq Fahran Nazar.

- There are plenty of jobs in finance, insurance, but also in many other industries, says Taariq. He has worked at Handelsbanken, Swedbank and Lynx before he started researching machine learning at the Department of Mathematics.

So why do you need to know mathematics to develop AI?

- First, you have to ask yourself what you mean by AI. It is a broad concept. As early as the 1950s, computer scientists began to describe systems that learn by themselves and that can exhibit something we interpret as intelligence. That intelligence is often formulated mathematically, since systems that learn are fundamentally based on mathematics.

- What we mean by AI today is actually what is called machine learning. Then we use data - for example, stock data or insurance data - to create systems that can make better decisions than us. All of that is done with mathematics. If you can't do math, you can't develop AI, says Taariq. He gives a simplified example:

Imagine that we have two pictures: one of a cat and one of a dog, but we are not allowed to see them. We feed them into a machine that will determine which is which. A very simple machine can only guess. A more advanced machine has “levers” – settings – that can be turned to make the result as good as possible. We either do it ourselves or we let mathematics do it for us.

– In reality, there are not two levers, but millions. Then it becomes impossible to adjust by hand, a mathematically defined way of adjusting the levers is required. That is why mathematics is the core. We cannot talk about AI without knowing mathematics, says Taariq.

– What you learn in mathematical statistics and machine learning is how to use mathematics to build systems and models: to analyze data, find patterns and be able to predict things that then become part of the decision-making basis.

- We are moving towards an increasingly data-driven world, where data is used to make things better. The skills we build here are the ones we will need in the future, says Taariq.

Read more about Mathematics and machine learning!