For more than a century, mathematicians and physicists have wrestled with the chaotic nature of fluid movement, such as how air swirls around the wings of an airplane or how water swirls inside a pipe. Google’s DeepMind Lab has recently made significant advances in this field using artificial intelligence.
With investors and others questioning whether AI is worth the astronomical costs, it’s heartening to see DeepMind working on something important like this. This is an example of AI creating something of real value.
As a reminder, DeepMind is a pioneering AI research organization acquired by Google over a decade ago. The organization is led by Demis Hassabis, a math and gaming whiz who quickly rose through the ranks at Google as AI became more powerful and important.
Demis Hassabis, Google DeepMind CEO Dan Kitwood/Getty Images
I got a “G” (yes, you read that right) in physics my freshman year of high school. To explain the significance of DeepMind’s recent discoveries, I turned to my daughter Nora Woolley, who is studying mechanical engineering and fluid mechanics. — At the University of Washington.
I shared the DeepMind blog and research paper with Nora, and she responded with her main points and some explanations for those struggling with math and physics.
“This could have huge implications for fluid mechanics and physics as a whole,” Nora told me.
please concentrate on this part
Let’s dive into the details with Nora’s help.
Fluids are so unpredictable that it is impossible to completely solve the equations used to model their behavior. To use these equations, physicists must make assumptions such as that viscosity is constant and that pressure varies smoothly.
Even simple scenarios can lead to an “explosion” where the equation predicts extreme outcomes such as infinite pressure or an incredible increase in velocity. These are called singularities and represent the moments at which mathematics can no longer predict the physical behavior of the fluid.
A singularity can be either stable or unstable. Stable singularities are easy to find, but unstable singularities are much more difficult to identify.
unstable genius
DeepMind researchers have discovered a new family of unstable singularities across three different fluid dynamics equations using machine learning and a bespoke physics-focused AI model.
By embedding the structure of the equations directly into these specialized AI models and incrementally optimizing them, the team achieved near-machine-level accuracy that allowed mathematicians to formally verify the results.
“This work provides a new strategy to address a long-standing challenge in mathematical physics,” DeepMind researchers said in their research paper. “This breakthrough represents a new way to conduct mathematical research,” the accompanying blog reads.
why is it important
The discovery of these new unstable singularities may help scientists better understand how turbulence, the unpredictable and energy-consuming behavior of fluids, occurs in nature and engineering.
This allows for a deeper understanding of areas such as aircraft drag, weather systems, blood flow, and energy distribution. These findings may make future flights less unstable.
Nora said the breakthrough could help monitor turbidity, a difficult-to-predict condition in fluids that is governed by momentum rather than physical properties.
“A lot of the software we use to monitor turbidity assumes that these equations are perfectly accurate across all values,” she told me.
DeepMind’s discovery of new unstable singularities may allow scientists to better monitor turbidity currents, she added, “because we now have a better understanding of the range over which these equations are valid.”
This doesn’t mean AI will cure cancer, but it’s much better than the generative AI slop that’s currently infesting the internet.
Sign up for BI’s Tech Memo newsletter here. Please contact us by email. abarr@businessinsider.com.
