Key points
- Complex equations once unsolvable now cracked using AI models
- AI tools accelerate breakthroughs in physics, chemistry, climate science
- Applications include drug discovery, battery tech, and protein folding
- Ethical concerns and data scarcity remain key challenges
ISLAMABAD: For decades, complex maths has been a major barrier in science, slowing breakthroughs in physics, chemistry, and climate research. Now, artificial intelligence is changing that—cutting down problem-solving times from years to mere minutes.
What once took scientists years of computing power or clever hacks can now be tackled in an afternoon, according to a new 500-page review co-authored by Dr Shuiwang Ji of Texas A&M University. The paper explores how AI is accelerating progress across the natural sciences.
It can take years for humans to solve complex scientific problems. With AI, it can take a fraction of the time.
Dr. Shuiwang Ji, a professor in the Department of Computer Science and Engineering at Texas A&M University and a leading expert in the emerging field of AI for… pic.twitter.com/8UcAZmJ14j
— Wevolver (@WevolverApp) July 29, 2025
At the heart of science lie differential equations, governing everything from quantum mechanics to weather systems. Solving Schrödinger’s equation, for example, is manageable for two electrons but becomes near-impossible for systems with millions of particles. As complexity increases, so does the number of variables, locking up even today’s fastest supercomputers.
Slow maths, stalled science
Traditional methods approximate solutions, but they require weeks of high-performance computing and often produce limited accuracy. This has long held back fields like quantum chemistry, vital for battery and catalyst development.
AI models, trained on existing solutions, can now predict results in seconds. These models—often based on graph neural networks—are designed to reflect physical laws like symmetry and conservation, making them both fast and reliable.
Density Functional Theory (DFT) is a quantum mechanical modelling method used to figure out the electronic structure of many‑body systems (atoms, molecules, and solids) by expressing the total energy as a functional of the electron density ρ(r).
Rather than solving the full… pic.twitter.com/b6Ctz61nA0
— Physics In History (@PhysInHistory) July 18, 2025
One such model recently replicated a month-long density functional theory (DFT) calculation in just ten minutes on a standard laptop.
These tools are already delivering results. AlphaFold, released in 2021, predicted the 3D structure of over 200 million proteins in a year—revolutionising biology.
Screening battery materials
Materials scientists are using AI to screen battery materials, while climate researchers are using neural solvers to cut computing costs by 40 per cent without losing accuracy.
What’s the potential and the risk of using AI for climate action?
📘 A new paper explores how AI can help cut emissions, improve disaster preparedness, and strengthen resilience — while also warning of risks like bias, data gaps, and high resource use.https://t.co/iKkBKqqPXr pic.twitter.com/Ys8lo2cSp0
— UN Climate Change (@UNFCCC) July 14, 2025
Still, challenges remain. Data is often scarce for frontier problems like fusion energy or rare-earth materials. Ethical risks also arise: the same AI that speeds drug discovery could aid in creating dangerous compounds. Researchers are pushing for safeguards, like screening molecules against threat databases before they are synthesised.
The paper, written by scientists from over 15 institutions, highlights the collaborative effort needed to tackle these issues. Ji’s RAISE Initiative links over 85 experts across disciplines, encouraging open data, shared tools, and joint research.
With AI becoming a standard tool in labs, scientists are shifting from solving equations to asking deeper questions—potentially changing how discoveries are made, and who gets to make them.
