We bridge the gap between nineteenth-century physics and twenty-first-century AI — one diffusion concept at a time.
Diffusion is one of the most fundamental processes in nature. It governs how a drop of ink spreads through water, how oxygen enters your bloodstream, how carbon atoms move through steel during heat treatment, and — surprisingly — how artificial intelligence generates images from noise.
Yet despite its ubiquity, diffusion is often taught in silos. Physicists learn Fick's laws. Biologists learn osmosis. Machine learning engineers learn denoising score matching. Rarely do these worlds connect.
Diffusion Science exists to change that. We write explainers that treat diffusion as a single, unifying idea — one that appears across physics, chemistry, biology, and artificial intelligence, connected by the same underlying mathematics.
Our goal is precision without jargon. Every article is written for a curious reader who doesn't have a PhD but isn't afraid of a little math. We define our terms. We use analogies carefully. We link concepts together so the big picture emerges naturally.
Diffusion models — the technology behind Stable Diffusion, DALL-E, and Midjourney — are named after physical diffusion for a reason. We explain the architecture, the training process, and the mathematics in accessible terms.
Brownian motion, Fick's laws, the Fokker-Planck equation. The classical physics of diffusion provides the mathematical foundation that AI researchers borrowed — sometimes knowingly, sometimes not.
From concentration gradients in a beaker to ion channels in a cell membrane, diffusion is the engine of life and chemistry. We explain how and why.
How we write about science
We assume intelligence, not prior expertise. We define jargon on first use and avoid hand-waving over the hard parts.
Every claim is backed by established science. We distinguish between settled consensus and active research.
We cross-link concepts across disciplines. The math of Brownian motion should illuminate the math of AI image generation.