One of my most joyful moments was while working on a high school assignment. The task was to create a flowchart detailing the possible decisions a character makes throughout a game. Given that this was a strict, all-girls Catholic school - where true happiness was confined to sewing the patterns of your husband’s boxers - the submission of my assignment resulted in an afternoon detention.

These afternoons were like any other day for me, so detention did not bother me. What stayed with me was the sense of completeness I felt during the assignment. It was probably the only time my chaos didn’t end in an unfavourable outcome. At that time, I asked my teacher, "Is there a place in this world where people construct these cobwebs of probabilities?" She suggested coding.

To me, coding algorithms is not just about following pseudocode; the challenge lies in ensuring computational efficiency. It's like following a baking recipe. As you bake, you try to use fewer utensils to save on cleanup time. This was fun, but it didn't feel quite right. With this pursuit on pause, I shifted my focus towards mathematics.

Flow chart of options to begin a human life.

My favourite area of mathematics has always been algebra, but I couldn’t love it enough to enter the realms of modern mathematics like category theory, number theory and so on. Then there are differential equations, the backbone of statistics and time. I am also a huge fan of geometry. Reading a book on geometry is like reading a history book, just with more numbers, funny-looking polygons and fewer evil people. The theorems in this topic are among the oldest in mathematics, dating back to Pythagoras and even before him.

I can't love these concepts enough to lose myself in them, but I also cannot dislike them enough to ignore them.

There must be a point where they all meet. I started my search by studying what I love most: algebra. In my spare time, I played around with matrix operations, and at rest, my mind often wandered off to projection matrices. After a stretch of self-doubt, I was reassured by a gem I found on page 2 of YouTube.

My favourite Mathematician, Gilbert Strang (MIT OpenCourseWare) ↗

I extended this by exploring the role of eigenvalues and eigenvectors. It became apparent that a linear regression line is built from the data points’ projections, or shadows. From here, I entered the world of AI, where I ran into the complications of scarce data. That wasn’t the main problem, though. The real problem was that predicting even the simplest thing, like tomorrow's weather, can never rely on one machine learning model but many, like a spider web within a bigger spider web: a cobweb of probabilities. But for ideas to blossom, you need a playground and an immense amount of patience.

I still have much to learn, especially in geometric analysis. Perhaps I need to revisit some books first to grasp the full potential of AI. After that, no barrier, not even Mother Teresa, can stop me.