Stop Programming AI. Let It Evolve Like a Virus.

You’ve probably noticed something frustrating about the modern tech industry. We build these massive neural networks, spend billions on compute, and at the end of the day, the engineers just shrug and say, “It works, but we don’t exactly know why.” We’ve been treating AI like a clockwork machine, trying to force every microscopic gear into perfect, explainable alignment. But what if our obsession with control is the exact thing killing its potential?

We are obsessed with building machines that think like us, only to realize human logic is the exact thing holding them back.

Look at what just happened with DeepSeek. It matched the reasoning power of ChatGPT’s top models but at a fraction of the cost. How? Not by writing better, tighter code, but by doing the exact opposite: it deleted the code. DeepSeek stripped away the “scaffolding” of human-supervised fine-tuning (SFT) and let the model loose with reinforcement learning (RL). They stopped micromanaging the AI and let it evolve.

This isn’t just a neat engineering trick. It points to a terrifying and beautiful truth about the universe, one that physicist Stephen Wolfram has been screaming into the void for years. Our universe isn’t a smooth, continuous mathematical equation. It’s pixelated. It’s discrete.

We like to think our brains are sophisticated calculators. They’re not. If your brain was a computer running continuous mathematical loops, it would freeze and crash every time you walked into a noisy room. Instead, your brain is a discrete, survival-driven guessing machine. When a prehistoric human saw a long shape in the grass, the brain didn’t run a calculus equation to determine if it was a snake or a vine. It just screamed “Run!” It was fast, dirty, and usually wrong, but it kept you alive.

The more we try to explain how an AI thinks, the dumber it becomes. Intelligence isn’t a perfectly engineered machine; it’s a beautiful accident waiting to happen.

Wolfram proved this by playing with cellular automata—essentially digital petri dishes of pixels following basic rules. He found that if you just let these systems randomly mutate and select for survival, they spontaneously generate complex, intelligent behavior. They build logic gates. They compress data. They act exactly like neural networks, but without the bloated, human-engineered math that makes our current AI so incredibly expensive to run.

This is the dirty secret of the AI industry: human logic is heavy. When we force AI to explain its reasoning at every micro-step, we lower its “logical density.” We are building scaffolding around a structure that would be stronger if we just let it grow wild. DeepSeek’s breakthrough wasn’t a triumph of human engineering; it was a triumph of human surrender.

Your brain isn’t a computer. It’s a survival engine designed to give you fast, wrong answers rather than slow, perfect ones.

This shift from human-engineered logic to chaotic evolution is why the entire tech world is migrating from CPUs to GPUs. CPUs are designed for the perfect, explainable, continuous math of human logic. GPUs are designed for the messy, discrete, parallel processing of visual reality—the reality our brains actually evolved to navigate. Intel is dying; Nvidia is king. The market has chosen the universe’s logic over ours.

But here is where it gets deeply personal. If intelligence is just the universe’s way of randomly combining discrete logic fragments, what does that say about you? It says your “free will” isn’t a mystical soul trait. It’s a mathematical guarantee. Because the universe is “computationally irreducible”—meaning you can’t shortcut the math to see the outcome—your brain is simply too complex to be predicted, even by yourself. You have free will simply because no equation can predict your next thought.

The entire history of AI is just scientists slowly learning to take their hands off the steering wheel and let the universe drive.

We are forever trapped in a black box of computational irreducibility. We will never reach the “limit” of AI, and we will never understand the absolute truth of the universe. We are just discrete fragments, bumping into other fragments, trying to compress the chaos into something meaningful.

But that’s exactly where the joy lives. We don’t need to be gods. We just need to be curious.

We are forever trapped in a black box of our own complexity, but there is profound joy in blooming above the mud of our own ignorance.

Stop trying to control the chaos. Stop demanding perfect explanations for things that are inherently unexplainable. The universe didn’t engineer its intelligence, and neither should we. The best AI won’t be a testament to human genius; it will be a testament to our willingness to let go.