AI for science: 5 ways it’s helping solve big challenges — from the lab to the field
AI isn’t just writing poems or suggesting meal plans anymore — it’s bringing new possibilities for science and what we know about the world.
Scientists can now decode electrons, create new materials and even “talk” to trees. Generative AI tools are speeding up the pace of discovery and unlocking insights about everything from the cells in our bodies to the ecosystems that sustain them.
“Scientific discovery is one of the most important applications of AI,” says Peter Lee, Ph.D., head of Microsoft Research. “We believe the ability of generative AI to learn the language of humans is equally matched by its ability to learn the languages of nature, including molecules, crystals, genomes and proteins.”
In the first half of 2025, Microsoft published numerous research papers in peer-reviewed journals and introduced new tools and collaborations across fields such as medicine, energy, biology and quantum physics. The goal: to accelerate how scientists explore complex questions and translate their findings into real-world impact using AI that’s powerful, practical and trustworthy, Lee says.
Here are five areas where AI is already making a tangible difference — and where the next breakthroughs may be just around the corner.
Health: Advancing care and research
AI is emerging as a vital partner in healthcare, not just for automating tasks, but for helping clinicians and researchers see more, understand faster and act earlier. From clinical notes to pathology slides, these multimodal models analyze large, unstructured datasets to spot patterns that help detect diseases and guide more tailored treatments.
One example is PadChest-GR,a first-of-its-kind dataset of 4,555 chest X-rays with pinpointed findings in Spanish and English. Developed by the University of Alicante and Microsoft, it can help radiologists interpret images more accurately and train AI models that learn and improve alongside scientists.
Another is the new Microsoft AI Diagnostic Orchestrator(MAI-DxO),which emulates a team of doctors by reasoning over multiple data sources. This research shows how AI could help tackle tough medical cases with higher accuracy and lower costs.
They’re part of a wave of science-first AI tools in healthcare, including GigaPath,which analyzes pathology slides at massive scale, and a project in Kenya to help prevent childhood malnutrition by identifying at-risk communities.
Discovery: Faster scientific insight
AI is helping scientists speed up research by analyzing complex data and simulating natural processes at a scale and pace that would be impossible otherwise.
Microsoft Discoveryis a new platform built with what’s known as agentic AI — systems that can reason, plan and take action, with permission — to act like a research teammate and automate tasks such as forming hypotheses, running simulations and refining experiments. It can recognize patterns and connections across large datasets, helping scientists test ideas more efficiently. In one early example, Discovery helped researchers identify a new datacenter coolant prototype in just over a week — a process that typically would have taken months.
Microsoft’s new AI model for Density Functional Theory(DFT)is helping solve a 60-year challenge in materials science by quickly and accurately simulating how electrons behave, which could help across applications from drugs to batteries and green fertilizers. Other tools like BioEmu-1,which helps decode protein structures, and MatterGen,which supports the development of new materials, are giving researchers more powerful ways to investigate and innovate.
Earth: New tools for a changing world
AI is moving from theory to real-world application as it helps scientists better understand the Earth’s complex systems and address environmental challenges.
Microsoft’s Aurora model is one of the first AI foundation models trained on Earth science data. It goes beyond weather forecasting to model how the atmosphere, land and oceans interact, helping scientists anticipate events like cyclones, air quality shifts and ocean waves with greater accuracy so communities can prepare for environmental disasters and adapt to climate change.
Other projects are applying AI to sustainability challenges in new ways. Researchers from Microsoft and the University of Washington are developinga low-carbon cement by mixing in seaweed biomass, creating a more sustainable building material. Avanade’s Intelligent Garden app “talks” to urban trees using sensors to monitor data like moisture, air quality and growth patterns and translating it all into a comprehensive health report. And in Tanzania, AI is helping conservationists track and protectendangered giraffes by analyzing drone footage and identifying individual animals through their spot patterns.
Quantum: Simulating nature
Quantum computing is expanding what’s possible in scientific research by simulating the natural world in ways conventional computers can’t. Traditional systems process information as ones and zeros, but quantum computers use quantum bits, or qubits, which can represent multiple values at once. That allows them to explore many possibilities simultaneously, making them especially useful for modeling complex systems like chemical reactions or material behavior.
Microsoft is combining quantum physics with AI to advance that kind of research. A recent breakthrough introduced 4D geometric codes,a new method for correcting errors in quantum hardware that makes it more stable, reliable and accessible. The company is also working with Atom Computing on a system using neutral-atom qubits, and its Majorana 1 chip represents an alternative quantum architecture to produce more reliable and scalable qubits.
Innovations like those are giving researchers new methods to model problems in areas where classical computing hits its limits, such as health, materials and climate.
Energy: Smarter, cleaner power
AI is playing a growing role in how we produce, store and use energy, by both optimizing current systems and helping build new ones.
Microsoft worked with Nissan Motor Corporation, for example, on a machine learning method that accurately predicts electric-vehicle battery wear — minimizing the need for lengthy physical testing — to help determine which batteries can be recycled instead of discarded. It’s an important part of Nissan’s initiative to reduce carbon emissions.
AI is also accelerating the development of nuclear fusion energy, a long-term goal for clean power. By simulating complex physical processes, scientists are testing ideas faster and identifying promising reactor designs to bring this energy to the grid sooner. In the U.S., Microsoft is exploring how AI can help streamline thepermitting process for advanced nuclear and fusion projects, which often face regulatory delays.
And in a step toward more sustainable energy storage, Microsoft used AI to screen more than 32 million candidates to discover a new material that could that could reduce lithium use in batteries by up to 70%.
