Research at Microsoft 2024: Meeting the challenge of a changing world

ViSNet emerged as a versatile tool capable of giving insight into the intricate relationships between molecular structure and biological activity.

Can how we think about our thinking help us better incorporate generative AI into our lives and work? Microsoft researchers explored metacognition’s potential to improve the tech’s usability.

This year, AI is expected to become more accessible, nuanced, and integrated in technologies that improve everyday tasks and help solve some of the world’s most challenging problems.

By training Orca-Math on a small dataset of 200,000 math problems, we have achieved performance levels that rival or surpass those of much larger models.

We hope to enable the developer community to benefit from the cache-store system’s performance gains and capabilities, to build on our work, and to expand the Garnet ecosystem by adding new API calls and features.

Leading researchers at Microsoft discussed how AI is transforming health care and the natural sciences, the intersection of AI and society, and the continuing evolution of foundational AI technologies.

Senior Researcher Chang Liu discusses M-OFDFT, which leverages deep learning to help identify molecular properties in a way that minimizes the tradeoff between accuracy and efficiency, work with the potential to benefit drug and materials discovery.

Principal Researcher Ida Momennejad brings her expertise in cognitive neuroscience and computer science to this in-depth conversation about general intelligence and what the evolution of the brain across species can teach us about building AI.

SAMMO streamlined the optimization of prompts, particularly those that combine different types of structural information, to enable AI practitioners and researchers to efficiently refine prompts with little manual effort.

Researchers proposed a method for leveraging the motion of small satellites to facilitate efficient communication between a large IoT-satellite constellation and devices on Earth within a limited spectrum.

NSDI provides a platform for researchers and experts to share insights, present research findings, and collaborate on the latest advances in the design, implementation, and evaluation of networked and distributed systems.

ASPLOS is the main forum where researchers bridge the gap between architecture, programming languages, and operating systems to advance the state of the art.

The new series “Ideas” debuts with guest Kalika Bali. The speech and language tech researcher talks sci-fi and its impact on her career, the design thinking philosophy behind her research, and the “outrageous idea” she had to work with low-resource languages.

Breakthroughs in robotics technology are poised to drive productivity, efficiency, and innovation across numerous industries. RASCAL improves the efficiency of vertical storage systems by operating across evenly spaced, parallel shelves and horizontal rails.

The confluence of digital transformation in biomedicine and the current generative AI revolution creates an unprecedented opportunity for drastically accelerating progress in precision health. Prov-GigaPath attains state-of-the-art performance on standard cancer classification and pathomics tasks, as well as vision-language tasks.

LoftQ adapts pre-trained language models to perform well in specialized applications. During fine-tuning, the model undergoes additional training on a smaller, task-specific dataset for improved performance.

FastGen optimizes the way LLMs store and access data, potentially cutting memory use by half while preserving their efficiency.

AutoGen Studio is the next step forward in enabling developers to advance the multi-agent paradigm.

Leading researchers at Microsoft dove into the importance of globally inclusive and equitable AI, shared updates on AutoGen and MatterGen, explored novel use cases for AI, and more.

The conference brings together experts from a wide range of disciplines who are committed to the responsible development of computational systems.

This conference covered a broad spectrum of topics, including 3D reconstruction and modeling, action and motion analysis, video and image processing, synthetic data generation, neural networks, and more.

Intelligence Toolkit is a human rights technology developed with global experts in the anti-trafficking community, yet applicable to a broad class of problems impacting societal resilience as a whole.

Trace is a new AutoDiff-like tool for training AI systems without using gradients. This generalization is made possible by OPTO, which can describe end-to-end optimization of AI systems with general feedback.

Researchers introduced an automated multi-agent framework for creating diverse, high-quality synthetic data at scale for language model post-training.

In an era increasingly steered by data, machine learning is a pivotal force, transforming vast amounts of information into actionable intelligence with unprecedented speed and accuracy.

With the skilled input of experienced game designers, tools like GENEVA could increasingly contribute to creating engaging gameplay experiences.

A multidisciplinary research team is exploring one solution: a credential that allows people to show they’re not bots without sharing identifying information.

AI has not yet delivered its full economic potential. Researchers at Microsoft are working to address the challenges that hold back progress.

MedFuzz is an adversarial machine learning method designed to reveal how much benchmark performance relies on unrealistic assumptions.

Eureka is an open-source framework for standardizing evaluations of large foundation models beyond single-score reporting and rankings.

College freshman Dexter Greene and Microsoft research manager Richard Black discuss how technology that stores data in glass is supporting students as they expand earlier efforts to communicate what it means to be human to extraterrestrials.

Leading researchers at Microsoft shared the latest multimodal AI models, advanced benchmarks for AI evaluation and model self-improvement, and an entirely new kind of computer for AI inference and hard optimization.

DRIFT Search introduces a new approach to local search queries by including community information in the search process.

The combination of Novo Nordisk’s industry-leading domain expertise and Microsoft’s industry-leading applied AI expertise is opening new and exciting possibilities to shape the future of life sciences.

Microsoft researchers are working to apply foundation models—large-scale models that take advantage of recent AI advances—to scientific disciplines.

Peter Lee discusses what’s next for AI in this GeekWire podcast. Peter’s comments begin at 29:20.

Since the early 1990s, the promise of AI has been a driving force at Microsoft Research, which has a track record of breakthroughs that continue to advance the state of the art.

Microsoft Research has been working on the development of efficient methods aiming for ab initio accuracy simulations of biomolecules. This method, AI²BMD (AI-based ab initio biomolecular dynamics system), published in the journal Nature, represents the culmination of a four-year research endeavor.

By unifying object recognition, detection, and segmentation into a single framework, BiomedParse allows users to specify what they’re looking for through a simple, natural-language prompt.

Researchers Chris Hawblitzel and Jay Lorch share how progress in programming languages and verification approaches are bringing bug-free software within reach. Their work on the Rust verification tool Verus won the Distinguished Artifact Award at SOSP ’24.

TamGen offers a new approach to drug discovery by applying the principles of generative AI to molecular design.

In an age where digital infrastructure underpins nearly every facet of modern life, SOSP serves as an important forum for showcasing the technologies that shape our interconnected world.

Research manager Karin Strauss and members of the DNA Data Storage Project reflect on the path to developing a synthetic DNA–based system for archival data storage, including the recent open-source release of its most powerful algorithm for DNA error correction.

Learn about Phi-4, the latest small language model in Phi family, that offers high-quality results at a small size (14B parameters).

More than 100 papers by Microsoft researchers and collaborators have been accepted at NeurIPS 2024, including five oral presentations and 19 spotlight sessions.

PromptWizard from Microsoft Research is now open source. It is designed to automate and simplify AI prompt optimization, combining iterative LLM feedback with efficient exploration and refinement techniques to create highly effective prompts in minutes.

As the “biggest election year in history” comes to an end, researchers Madeleine Daepp and Robert Osazuwa Ness and Democracy Forward GM Ginny Badanes discuss AI’s impact on democracy, including Daepp and Ness’s research into the tech’s use in Taiwan and India.

According to Ashley Llorens, corporate vice president and managing director at Microsoft Research, AI models will soon be able to handle far more complex tasks.