←
Eric Horvitz3,059 posts
Eric Horvitz avatar

Eric Horvitz

@erichorvitz

Chief Scientific Officer, Microsoft

◷ Joined March 2011▧ 283 Photos and videos
Eric Horvitz avatar
Eric Horvitz@erichorvitz·34m

Leveraging "capability overhang" Strong & prescient work over the years on GraphRAG. GraphRAG Zero enables AI agents to access rich sources of knowledge, lighting them up with enterprise- and domain-specific knowledge bases. @MSFTResearch

GraphRAG Zero solves a fundamental challenge that has limited AI agents' ability to reach their full potential. It’s a technology for holistic discovery that reflects years of Microsoft research, transforming graph-based retrieval from an experimental technique into a deployable system. Optimized for real-world applications and agents, it enables organizations to explore connections across vast amounts of their own data, quickly and at low cost. It’s available now in Foundry IQ and @github’s Repo Mind Light, supporting applications from scientific discovery to repository understanding. msft.it/6017xE5xL

1:10
Eric Horvitz avatar
Eric Horvitz@erichorvitz·52m

Can AI help build a healthier future? The Shape of Things to Come podcast Enjoyed the conversation with @dcburger, @avapamini, and @EricTopol. @MSFTResearch

AI is opening new possibilities in medical research. But what will it take to turn scientific breakthroughs into better health outcomes? In the latest episode of The Shape of Things to Come, Microsoft Research Technical Fellow and CVP @dcburger brings together: @avapamini, Principal Researcher Manager at Microsoft Research, @erichorvitz, Chief Scientific Officer at Microsoft, and Dr. @EricTopol, Executive Vice President, Scripps Research to explore how AI is reshaping our understanding of biology and medical care and what it could mean for the future of medicine. From accelerating scientific discovery to the challenges of real-world experimentation and clinical implementation, the group examines both the promise of AI and the work still ahead. Could these advances help shift medicine toward earlier intervention and prevention? Watch or listen to the podcast: msft.it/6019x1y5X

2:28
Eric Horvitz avatar
Eric Horvitz@erichorvitz·Nov 11

Streaming intelligence Always on, always reflecting, always thinking ahead. Lots of headroom to press these principles into service today. The top-level ideas around "streaming Intelligence" are growing in importance. The work aimed at identifying timeless principles of intelligence. Agents that do this well will be the smartest & most productive agents across the ecosystem. Thanks @tomssilver Paper: erichorvitz.com/cc_aij_horvi… @MSFTResearch @DimitrisPapail @vkontonis @yzeng58

Post attachment

This week's #PaperILike is "Principles and Applications of Continual Computation" (Eric Horvitz [yes, that one!], 2001). An early work on "sleep time" compute, increasingly relevant for agents (e.g., anthropic.com/research/compo…). PDF: erichorvitz.com/cc_aij_horvi…

Eric Horvitz avatar
Eric Horvitz@erichorvitz·Nov 11

Discovery, study, and coordinated response The methodology is as important as the finding. We have to stay ahead of potential cybersecurity risks by safely pressing AI models into service to help us better understand & mitigate the gaps and possibilities. Paper: microsoft.com/en-us/research… Microsoft Research blog: microsoft.com/en-us/research… MSTIC guidance: microsoft.com/en-us/security… @igortsyg @MarkRussinovich @DimitrisPapail @kptales @blakebullwinkel @ram_ssk @satyanadella

Dimitris Papailiopoulos @DimitrisPapail Nov 10 Article Lazare, veni foras: Reprogramming Dormant Neurons to Backdoor AI Agents We find that weight reprogramming fewer than 10 dormant neurons in an LLM is enough to backdoor any agent built on it, with no training required. The model passes evals and behaves normally until a

Eric Horvitz avatar
Eric Horvitz@erichorvitz·Nov 10

We launched AI100 as we reached +5 years after the 2008-2009 Presidential Panel on Long-term AI Futures, that I organized while president of AAAI (the AAAI Asilomar Study), during a presidency I themed "AI in the Open World." In 2013, it seemed we needed to have another study. But that would always be the case: Why not do such a study on AI every 5 years--forever? Here a link to on the Asilomar study (2009): aaai.org/about-aaai/aaai-pre…

Eric Horvitz avatar
Eric Horvitz@erichorvitz·Nov 10

The 2026 report of the One Hundred Year Study on AI (AI100) was published today. The project is endowed to continue to observe, assess, and provide reflections & guidance on AI and its influences over the lifespan of Stanford University. We expect this to be hundreds of years. @Stanford @SheilaMcIlraith @conitzer @Rbaltman @mlittmancs @PeterStone_TX @BarbaraGrosz @DMulliganUCB @tommmitchell @AlanMackworth @yshoham ai100.stanford.edu

Eric Horvitz avatar
Eric Horvitz@erichorvitz·Oct 25

Engaging discussion. Recorded earlier in the summer before the recent mathematics results.

AoT@AoT_WEOct 25

Scientists solved about 180,000 protein structures in 60 years. AlphaFold has predicted 200 million+. @Microsoft's chief scientist @erichorvitz says science is changing so fast it's rethinking what integrity and credit even mean.

1:21
AoT avatar
⟳ Eric Horvitz retweeted
AoT@AoT_WE·Oct 24

It feels like only the AI labs and governments know where this is going. @Microsoft's chief scientist @erichorvitz says it's uncomfortable to live through a transformative era, and there's a lot of uncertainty ahead. Even for them.

0:24
Brad Smith avatar
⟳ Eric Horvitz retweeted
Brad Smith@BradSmi·Oct 22

If we believe AI is one of the most powerful technologies we’ve ever created, we should also recognize that no technology is above the law. Innovation needs to keep moving forward. But safety needs to keep pace. We need sensible, balanced rules that enable us to do both.

2:24
Stanford Engineering avatar
⟳ Eric Horvitz retweeted
Stanford Engineering@StanfordEng·Oct 23

“We don’t say, how can we use an existing tool well? We look at the problem and say, how can we make a new tool that will allow us to answer questions much, much more effectively?” It’s how @KarlDeisseroth and his colleagues developed optogenetics.

0:42
Eric Horvitz avatar
Eric Horvitz@erichorvitz·Oct 23

Enjoyed the conversation with these vibrant hosts. @AoT_WE

AoT@AoT_WEOct 23

Britain's railway boom hit 7.3% of GDP in 1847. Shares then fell about two thirds, but the lines kept running. Microsoft's chief scientist Eric Horvitz expects AI to follow the pattern: some investors lose heavily, the infrastructure wins.

1:04
Science Magazine avatar
⟳ Eric Horvitz retweeted
Science Magazine@ScienceMagazine·Oct 22

Researchers have developed an AI-based approach to help physicians with real-life clinical workups, starting with the patient’s chief complaint, systematically asking questions and deciding what laboratory and imaging tests to order, and then continuing the evaluation on the basis of the answers received. Learn more this week in Science: scim.ag/3U8xn94

Post attachment