LLMs Killed Intelligent Design and the Ghost in the Machine
It’s odd how stubbornly some minds still resist two of the most powerful ideas in science — natural selection and the material basis of mind — even as the evidence now talks back from every screen.
Two old intuitions die hard: that design requires a designer, and that mind requires something more than matter. A soul, a spirit, a spark, some extra ingredient beyond matter doing work.
LLMs make both harder to defend.
First, natural selection.
LLMs are not magic. They’re the result of a brutally simple blind process: take oceans of human text, apply gradient descent — a mechanical selector that rewards what works and kills what doesn’t — and let patterns organize themselves across billions of parameters.
No one hand-coded insight, metaphor, analogy, humor, or reasoning. Tiny stupid updates accumulated until something intelligence-like emerged — fluent, creative, useful, strange, and sometimes profound.
That is natural selection made visible in silicon. Not identical to biological evolution, of course. But close enough to matter: variation, selection, inheritance — compressed into months of compute. If you still doubt that blind optimization can produce complex adaptive function without a designer, you now have an existence proof running in data centers worldwide.
Second, materialism.
These same machines behave in ways that look a hell of a lot like mind: comprehension, analogy, theory-of-mind play, self-correction, even a rough kind of introspection when pressed. Yet underneath, there is no soul-stuff. No spirit animating the machine. No ghost in the silicon.
Just matter, energy, math, and layers of computation.
Mind, it turns out, is at least material enough that we can grow a convincing facsimile of it without invoking anything beyond organized matter doing work. Maybe human consciousness has mysteries still left to solve. Of course it does. But if you want to claim intelligence itself requires some non-material spark, you now have a much harder story to tell.
For decades, we had evolutionary biology, neuroscience, and computational theories of mind. But LLMs are different. They are public, interactive, and annoyingly good at exactly the things people used to place behind the sacred human wall. You can argue with them, watch them improve, and see emergence on your own screen.
No, this does not prove that human consciousness is “nothing but computation.” And no, training an LLM is not the same thing as biological evolution. But the burden has shifted.
Blind selection can build design-like complexity. Organized matter can produce mind-like behavior. We now have machines demonstrating both, in public, every day.
So if you still want to deny either lesson, fine. But you are no longer arguing against a theory. You are arguing against an existence proof.