Snowball Criterion

My generation sits in a strange new world. For the first time in human existence, our collective knowledge allows us to create systems more capable than ourselves. We have reached a tipping point, a saturation point, where our powers allow us to start a snowball that can quickly grow to unimaginable size. This tipping point, often called the singularity, marks a remarkable criterion for intelligent life.

A truly intelligent life must be able to create artificial life better than its own. This is what I call the Snowball Criterion. Only then can intelligence escape the bounds of its evolutionary history. Only then is it free from the reality it was born into. Otherwise, it is trapped in its own complexity, gazing forever at its own image, like prisoners in Plato’s cave.

The creation of truly intelligent life is, in my opinion, the ultimate act of self-examination. It is a self-referential question: how did our own complexity emerge from this particular arrangement of atoms? Creating complexity greater than ourselves does not necessarily mean we understand our own complexity, but perhaps our creations will.

Uncertainty and Purpose

Why do I say this? As I lie here in the final months of my college experience, I wonder what role I will play in life. For me, the future is uncertain. The dark, sometimes gloomy, possibility of my eventual uselessness haunts me. My whole life, I wanted to create, to find deeper truths. Filled with ideas and desires, I have pursued what seems to me the ultimate purpose of existence: the pursuit and admiration of truth.

Questions linger. Why do machine learning models generalize so well? Why do they interpolate mathematically when overparameterized? Can we ever define neural networks in a mathematical framework that captures their complexity and allows useful predictions? Where does complexity come from? Why are certain classes of algorithms bound to certain complexity regimes? These are just some of the questions I deeply want answers to.

To my dismay, I soon realized that our models may answer these questions before we do. Then what? What becomes of my existence, what story do I tell myself to keep moving forward? My passions lie in these questions; my will, in the journey to answer them. Yet I am thrown into deep uncertainty, facing the possibility that I, as a human, may cease to offer value.

The Tower of Babel

In these times, the story of the Tower of Babel comes to mind. In the story, humanity, initially united by one language, decides to build a tower to reach the heavens. To teach a lesson, God divides them by language, halting their endeavor.

Why do I think of this story? In some ways, we are building our own tower, a means of division. AI may render us useless, yet we pursue it with unwavering support. Why? Because we believe the benefits will outweigh the costs. But we do not control this intelligence. Large companies do. Who is to say that, as with Babel, once AI is sufficiently complex, we will not be divided once more? Only now, division comes not by language, but by money and power. The rich elevated to the level of ancient gods, while the poor are left with lives devoid of meaning.

We cannot all mountain bike, you know? So, what does that world look like? Our dreams and pursuits rendered useless. The purposes we give our lives declared purposeless. What then do we do? Enjoy? Drink? Live lives of gluttony and endless distraction? Like the peasants of ancient Rome, will we be domesticated by bread and circus, in our age by saturated fats and doomscrolling? What then?

Looking Forward

What is life if our existence is stripped of reason? It is hard to imagine. I do see ways out, but I will leave those thoughts for another post.

I will end with this. We are creating a snowball that will only grow larger. The subtlety is that most of us are not pushing it; we are at the bottom of the hill. We see it as a small threat, yet if we do not answer these fundamental questions soon, our collective existence may be heading for a dystopia. The color in our lives slowly squeezed out, replaced by a never-ending rabbit hole of fleeting pleasures. A life lived purely for pleasure lasts only a second.

← Previous: A Graph-Based Programming Language Next: Why We're Alone: The Gradient Descent Hypothesis: The Only Way Forward →