In their 2024 SOSP paper, researchers explore a common—though often undertested—software system issue: retry bugs. Research manager Shan Lu and PhD candidate Bogdan Stoica share how they’re combining traditional program analysis and LLMs to address the challenge.Read the paper

Every year, interns from academic institutions around the world apply and grow their knowledge as members of the research community at Microsoft. In this Microsoft Research Podcast series, these students join their internship supervisors to share their experience working alongside some of the leading researchers in their respective fields. In this episode, Angela Busheska, an undergraduate engineering student at Lafayette College, talks to Senior Researcher Vaishnavi Ranganathan, about her work on TerraTrace, a platform that brings together statistics and large language models to track land use over time for agricultural and forestry applications. Busheska discusses the personal loss that drew her to climate activism, the chain of events that led to a memorable face-to-face meeting with Microsoft’s chief sustainability officer, and her advice for going after the internship you want and making the experience count.Learn more:TerraTrace | GitHub repoProject FarmVibes | Project homepageProject FoodVibes | Project homepage

The personalizable object recognizer Find My Things was recently recognized for accessible design. Researcher Daniela Massiceti and software development engineer Martin Grayson talk about the research project’s origins and the tech advances making it possible.The Find My Things story is an example of research at Microsoft enhancing Microsoft products and services. To try the Find My Things tool, download the free, publicly available Seeing AI app.Learn more:Find My Things: Personalized Accessibility through Teachable AI for People who are Blind or Low Vision | Publication, May 2024Understanding Personalized Accessibility through Teachable AI: Designing and Evaluating Find My Things for People who are Blind or Low Vision | Publication, October 2023Teachable AI Experiences (Tai X) | Project pagePeopleLens | Publication, June 2021ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognition | Publication, October 2021Collaborators: Teachable AI with Cecily Morrison and Karolina Pakėnaitė | Microsoft Research Podcast, December 2023

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.Learn more:Avenues: The World School — Golden Record 2.0Project homepageGolden Record: OverviewNASA ScienceProject SilicaProject homepageSealed in glassMicrosoft Unlocked innovation story, 2023Optics for the cloud: storage in the zettabyte era with Dr. Ant Rowstron and Mark RussinovichMicrosoft Research Podcast, November 2019Project Silica proof of concept stores Warner Bros. ‘Superman’ movie on quartz glassMicrosoft Source blog, November 2019

Model maker and fabricator Lex Story helps bring research to life through prototyping. He discusses his take on failure; the encouragement and advice that has supported his pursuit of art and science; and the sabbatical that might inspire his next career move.Learn more:Microsoft PremonitionProject EclipseProject PRISM3D TelemedicineJacdacAudio Devices

In this episode, Microsoft Product Manager Shrey Jain and OpenAI Research Scientist Zoë Hitzig join host Amber Tingle to discuss “Personhood credentials: Artificial intelligence and the value of privacy-preserving tools to distinguish who is real online.” In their paper, Jain, Hitzig, and their coauthors describe how malicious actors can draw on increasingly advanced AI tools to carry out deception, making online deception harder to detect and more harmful. Bringing ideas from cryptography into AI policy conversations, they identify a possible mitigation: a credential that allows its holder to prove they’re a person––not a bot––without sharing any identifying information. This exploratory research reflects a broad range of collaborators from across industry, academia, and the civil sector specializing in areas such as security, digital identity, advocacy, and policy.

Researcher Brendan Lucier and professor Mert Demirer are applying their micro- and macroeconomic expertise, respectively, to forecasting the economic impact of AI. They share how they’re using a task-level breakdown of occupations to help predict the future.Learn more:AI, Cognition, and the Economy (AICE) | Initiative pageIdeas: Designing AI for people with Abigail Sellen | Microsoft Research Podcast, May 2024

Emre Kiciman shares how some keen observations and a desire to have front-end impact led him to make the jump from systems and networking to computational social science and now causal analysis and large-scale AI—and how systems thinking still impacts his work.Learn more:AI Controller Interface: Generative AI with a lightweight, LLM-integrated VM (blog)AICI: Prompts as (Wasm) Programs (GitHub)

A lack of appropriate data, decreased model performance, and other obstacles have made it difficult to expand the input language models can receive. Li Lyna Zhang introduces LongRoPE, a method capable of extending content windows to more than 2 million tokens.Read the paperGet the code

Senior Researcher Arindam Mitra introduces AgentInstruct. Using raw data sources, the automated multi-agent framework can create diverse, high-quality synthetic data at scale for the post-training of small and large language models.Read the paper