Empowering every scientist with AI-augmented scientific discovery

At Microsoft, our vision is to empower scientists with the latest breakthroughs in AI to unlock their full creative potential and tackle some of our most pressing challenges. This vision will require bringing the full power of generative AI together with quantum-classical hybrid computing to augment every stage of the scientific method. Whether expanding knowledge research, creating better hypotheses, or accelerating experimentation and analyses, doing so demands a purpose-built cloud platform for science. This is why we built Azure Quantum Elements for chemistry and materials science.

Today, we’re announcing Generative Chemistry and Accelerated DFT, which will expand the ways researchers can harness the full power of this platform. These breakthrough capabilities will empower scientists to compress the next 250 years of chemistry into the next 25.

With Generative Chemistry, we want to broaden the horizons of scientific exploration. Researchers can generate and explore novel molecules suited for specific industry applications using the latest AI models trained on hundreds of millions of compounds, and then evaluate the steps suggested by the workflow for synthesizing the most promising candidates in a lab more efficiently — all in a matter of days rather than years.

With Accelerated DFT, researchers can expedite and scale their chemical discovery pipelines by simulating the quantum-mechanical properties of molecules at an unprecedented speed — an order of magnitude faster compared to other Density Functional Theory (DFT) codes.

This brings us closer to a new paradigm for scientific discovery, where advanced AI and digital tools are more accessible than ever to scientists, students, and labs across industries. Below is our vision for how researchers will be able to leverage these breakthrough capabilities to design new molecules and enable the transformation of entire sectors from consumer goods and medicine, to manufacturing and energy, in turn addressing some of our most pressing societal challenges.

YouTube Video

We’re working towards this vision today. As part of the private preview of Azure Quantum Elements, scientists and developers have the opportunity to explore Accelerated DFT today, with the potential to access Generative Chemistry in the coming weeks.

We’re already putting our vision into practice by collaborating with Unilever, a global leader in consumer goods, which serves over 3.4 billion people every single day. Unilever is harnessing the power of Microsoft supercomputing and AI services to support their digital R&D transformation and product innovation.

Integrating AI into every stage of the scientific method

From global ambitions like reversing climate change and pioneering renewable energy sources to personal ones like living more sustainably and using healthier and safer products, we all want to do our part to create a better world. Time is of the essence for many of these goals, with more than 8 million scientists1 around the globe working to pioneer innovative solutions and unlock progress. At Microsoft, we aim to empower them with state-of-the-art digital tools to harness the full collective ingenuity of every researcher and lab around the world.

Just as generative AI has unleashed new waves of creativity and improved productivity with collaborative tools like Copilot, we are now bringing AI and natural language processing capabilities to science. Our goal is to integrate AI reasoning into every stage of the scientific method: this requires the power of next-generation AI models to speed up the scientific process from hypothesis to results. It starts with knowledge research and hypothesis generation, connecting the dots by generating millions of potential molecular candidate solutions, then narrowing down candidates with digital experiments and analyzing the outcomes — all in a matter of days. We demonstrated how this approach can land real-world results in our collaboration with PNNL, where we screened over 32 million candidates to discover and synthesize a new material that holds the potential for better batteries — a tangible example of the possibilities in this new era of scientific discovery.

When powered by natural language tools, this new paradigm will help create an autonomous reasoning loop with AI at every stage as a scientific assistant. It will redefine how we approach innovation by democratizing these capabilities for breakthrough discoveries.

A reasoning loop of AI processes
Our goal is to integrate AI into every stage of the scientific method, from initial research and hypothesis generation to experimentation and analysis.

Announcing new capabilities in Azure Quantum Elements

Generative Chemistry will unleash a new wave of creativity for scientists tasked with discovering and designing new molecules. This will enable breakthrough growth across many industries, whether helping an oil and gas company discover a stronger fuel additive for enhancing the longevity of engine life, or an adhesive firm creating a new chemical for strengthening adhesion while removing unwanted residue.

We could compare this discovery process to searching for a small box in a large, crowded and dark warehouse with one small flashlight. We can only focus the light on a small area at a time while the rest of the warehouse remains completely dark and unknown. Generative AI gives us a much smarter light that can point in new directions, providing visibility where we may not have considered – or have been able – to look before.

Researchers can ask Generative Chemistry for molecules with desired characteristics, such as the ability to degrade rapidly or be recycled more easily. They can also provide information about their targeted application and let the system help determine relevant molecular properties. After a few more steps, they receive a set of candidates — matching those parameters — for further study.

However, simply generating candidates is not sufficient for transforming the discovery process with AI. The essential criteria for computational tools in chemistry are that they help scientists discover molecules that are novel, synthesizable and useful in the real world. This is why I’m excited to see our approach to Generative Chemistry come to life, suggesting molecules that have not been seen before, with useful properties tuned for a specific application, and whose synthesis is feasible in a reasonable number of steps.

For this reason, Generative Chemistry will offer researchers potential steps to consider as they develop their “recipe” for synthesizing these molecular candidates in a laboratory. Support for this critical component has been developed from the capabilities of our AutoRXN software, exploring chemical reactions in reverse order, which can help to evaluate synthesis pathways for creating a target molecule.

Visualization of a funnel narrowing down molecules using AI to finish with best possible candidates.
After scientists specify the desired characteristics of a molecule, they receive thousands of molecular candidates that can be further refined through AI inferences and subsequent HPC simulations, arriving at a select few that show the most promise for laboratory synthesis and further experimental exploration.

This capability is truly groundbreaking for scientific discovery. Businesses and research groups can look for efficient, cost-effective and innovative methods to develop new molecules in a matter of days, compressing the iterative process of extensive database searches and trial-and-error laboratory experiments. This end-to-end workflow will provide scientists with entirely new compounds that could lead to the next breakthrough in manufacturing, medicine and more.

We’re also announcing Accelerated DFT to offer a simplified and more powerful quantum chemistry solution for scientists. For the past few decades, DFT has been an extremely popular method used across a variety of molecular simulations, helping researchers to simulate and study the electronic structure of atoms, molecules and nanoparticles, as well as surfaces and interfaces.

We can liken molecular systems to traffic systems, where cars moving in various directions at different speeds represent electrons. From a traffic helicopter, we can observe the overall flow of traffic even if we don’t know each car’s speed and destination. DFT provides this “helicopter view’” of molecular systems, simplifying the complex task of tracking individual electrons by instead mapping out the “density” of them at a higher altitude.

Such DFT simulations can be complex to optimize and run, and often require supercomputer-scale resources. This is why our managed DFT service, based on innovation developed by Microsoft Research, enables researchers to perform substantially faster calculations than other DFT codes and offers a 20-fold average increase in speed compared to PySCF, a widely used open-source DFT code.

Accelerated DFT is already used by many organizations such as AspenTech, DTU Energy University of Denmark and Unilever. It seamlessly integrates into broader chemistry and materials science workflows, and paves the way for expediting innovations in therapeutics, environmental sustainability and beyond.

You can learn more about this announcement in the technical blog, Introducing two powerful new capabilities in Azure Quantum Elements: Generative Chemistry and Accelerated DFT.

Pioneering a new scientific discovery paradigm with Unilever

Unilever stands at the forefront of the consumer goods industry, with a strong portfolio of household brands that are used by 3.4 billion people every day, including Dove, TRESemmé, Omo, Degree, Hellmann’s and Ben & Jerry’s. Whether cleaning, beauty or care products, each requires the latest scientific breakthroughs to ensure the best possible consumer experience and enhance daily life.

Over the past two and a half years, Unilever has worked with Microsoft to identify new digital capabilities to drive product innovation forward. Unilever is bringing its digital vision to life through the transformational DataLab — its digital counterpart to the company’s physical laboratories — with the help of Microsoft Azure. From unlocking the secrets of our skin’s microbiome to reducing the carbon footprint of a multi-billion-dollar business, Unilever is redefining what it means to be a consumer goods company in the modern world with leading science.

YouTube Video

With Copilot and the advanced simulation capabilities of Azure Quantum Elements, Unilever can query scientific information using natural language, performing thousands of computational simulations in the time it would take to run tens of laboratory experiments. Unilever scientists can use the data gathered from these simulations to fine-tune models that screen tens of thousands of materials at substantial speed or enable the exploration of intricate chemical reactions.

For example, R&D teams can expand their search space for novel molecules that restore natural bonds in hair fibers across more hair types, in turn redefining the standards of personalized hair care for brands like Dove and TRESemmé. Furthermore, by placing scaled simulations at the forefront of the discovery funnel, Unilever will be further empowered to expedite the delivery of solutions within their key sustainability focus areas.

“Digital tools are unlocking an unprecedented age of scientific discovery. Using advanced computing power and AI, we are able to compress decades of lab work into days, accessing a level of insight we could not previously have imagined. This technological leap, coupled with our vast repository of proprietary data and a century of expertise in personal and household care, means our scientists are able to lead the industry in developing the next generation of consumer goods.”
Alberto Prado, Global Head of R&D Digital and Partnerships at Unilever

Expanding quantum capabilities in Azure Quantum Elements

We stand on the cusp of unprecedented innovation, and at Microsoft, we continue to pioneer state-of-the-art solutions to usher in a new era of scientific discovery. We remain focused on achieving scaled quantum computing and more breakthroughs on our path to engineering our topological qubits with inherent hardware-level stability.

Earlier this year, we demonstrated with Quantinuum the most reliable logical qubits on record, further advancing the state-of-the-art for quantum computing. And recently, we simulated a chemical catalyst combining classical supercomputers, AI and logical qubits created with Microsoft’s qubit-virtualization system and Quantinuum’s H1 hardware. This combination holds the key to unlocking scientific breakthroughs enabled by a new generation of hybrid-computing applications.

In the coming months, we will bring advanced logical qubit capabilities using our software and Quantinuum’s hardware in private preview in Azure Quantum Elements. As logical qubit capabilities scale to deliver increasingly reliable results, we will unlock simulation accuracy, moving us from scientific advantage to commercial advantage, and ultimately to solving some of the world’s most pressing problems.

Accelerating scientific discovery, together

We’re committed to advancing these technologies responsibly, always focusing on innovation, empowerment and trust. That’s why we are committed to responsible computing practices and the Microsoft AI principles, to help ensure that safety measures adequately account for the increasing power of AI and quantum.

For more information about today’s announcements:

Top image: Leaders from Unilever and Microsoft discuss the Azure Quantum Elements program.


1. “Statistics and resources | 2021 Science Report.” This translates into 8.854 million full-time equivalent (FTE) researchers by 2018”.

The post Empowering every scientist with AI-augmented scientific discovery appeared first on The Official Microsoft Blog.

This post originally appeared on The Official Microsoft Blog.

Leave a Reply

Your email address will not be published. Required fields are marked *