We humans made significant breakthroughs in our understanding of Nature. However, most scientists would tell you that there is still more to be discovered than things we fully understand.

Many of our biggest discoveries were made via reductionism - studying simpler phenomena and using them to understand complex systems as the sum of its parts. A great example of this is classical mechanics. Each object has forces acting upon it, and studying these forces allows us to predict, for example, the motion of stars in the solar system.

What we humans are terrible at is complex systems for which reductionism isn’t that straightforward, as is the case for most biological systems, the brain, and particularly the human body.

Another example of this is Artificial Intelligence (AI). We were able to devise systems whose behavior is complex, and in many cases we cannot point out to why our AI systems behave the way they do. That’s why many people say “AI is a blackbox”. What we do know about AI though, is that it’s amazingly good at deciphering complex relationships. Given enough examples, computational power, and depth, we’ve seen AI describing photos with language, translating text, turning black & white videos to color, generating musical compositions, and predicting election results.

I believe that the secret to unlocking the behavior of complex systems is already available. Instead of approaching every problem with a reductionist mindset, we can leverage AI’s ability to find complex relationships for us. So take our most poorly understood complex systems like the human body or the brain, and use AI in order to find the patterns that guide its behavior.

What I’m not suggesting is using AI in order to predict the behavior of these complex systems. I’m suggesting to use AI to predict the mathematics needed to understand complex systems. In other words, AI should be significantly better at reductionism than we humans, so why not leverage it for that purpose?

In particular, the mathematics needed to understand optimal human nutrition would and should be derived by AI, who can capture all the complexities we often overlook.

One day, AI would be used to explain why AI makes the decisions it makes.