Every solution exists in the universe of all possible programs, we just have to search for them…
When you look into the world today, you can easily see the difference between objects created by humans and those that are created by nature. But what is this difference exactly and how would we define it? I think nature works by applying very simple rules, in a fractal sort of way, over and over. If you take a tree for example, you could say that every subtree in the tree has the same structure as the tree itself. This is fractality is enough to create very complex systems, like me writing this post.
Human engineering on the other hand is very different. We have very complex structures, that we combine with each other over and over, but you don’t see this fractality anywhere. A house does not contain smaller houses. It’s not scale invariant. It looks different when you zoom in.
I think the bigger question I want to answer here is, if we can call the fractal nature way of doing things intelligent or not. We can definitely call it computational though. So the question is this:
Intelligent != Computational ?
If we believe Stephen Wolfram, which I’m a big fan of, there is no distinct line between the intelligent and the computational. This is what he calls the Principle of Computational Equivalence. Here is how he defines it:
Most systems are computationally equivalent. For example, the workings of the human brain or the evolution of weather systems can, in principle, compute the same things as a computer. Computation is therefore simply a question of translating inputs and outputs from one system to another.
If I would put it in my own words I would describe it like this: If you have a system that is computationally universal, like a Turing Machine, you can generate all computational possible outcomes. This means that there is no inherent difference between the intelligent and the computational. Most systems in nature are, after the definition above, intelligent.
How does this help us in the real world?
If we want to solve a problem, we can simply search for a program in the universe of all possible programms. The problem is that this universe is incredibly large. If you are trying to brute-force search through it, you will very likely not find anything. But don’t we already have a smart way to search this program space?
Yes we do, it’s called Machine Learning. This new way of doing things or how Andrej Karpathy calls it Software 2.0, is just a clever way of searching in this program space. We will only need to define the goal we want to reach and then let the algorithm search the universe of all possible programs, basically nature, for a solution. This is why in the future things will look more like nature:
I think it is very ironic to see a shift back to nature, after we tried for so long to distance us from it. After millenia of human technological progress, we are just figuring out natures intelligence, power and efficiency. This future, where we let nature figure things out by itself, will open infinite possiblities.
Feature engineering is dead! Long live nature!
