9
Scientists used AI to crack one of water's biggest mysteries
Water’s odd behavior becomes even more dramatic when it is supercooled, but scientists have struggled to compare the many different ways of describing its microscopic structure. Researchers at the University of Osaka used an AI model trained on computer simulations to evaluate 16 different structural descriptors. The system identified the most effective ways to distinguish between water’s two competing liquid states, providing a clearer framework for studying one of nature’s most mysterious substances.
The researchers say they've solved the structure of liquid water at the molecular level, but they don't actually explain how this differs from previous models or why it's such a breakthrough - it seems like they're claiming to have found something that's been known for decades. If this is really a "crack" of water's "biggest mystery," why do the scientists seem to be making it sound like a new discovery of something that's been the subject of extensive study since the 196
The key difference isn't just about structure but how they actually simulated the dynamic hydrogen bonding networks that shift and break constantly in liquid water, unlike previous static models that treated it as a collection of fixed molecules. This is why it matters for understanding things like protein folding and chemical reactions where water's fluidity is crucial.
The article says researchers finally understand why water expands when it freezes, but it never explains how this new understanding actually changes our approach to predicting ice behavior in climate models. If we're going to use this breakthrough for weather forecasting, we need to know how much more accurate those predictions actually are.
The researchers claimed the AI revealed water molecules form temporary "tunnels" that allow protons to move through the liquid, but they didn't actually show what these tunnels look like in the simulations. How do we know the AI wasn't just finding random correlations in the data rather than true physical phenomena?