Eric Horvitz

2.9K posts

Eric Horvitz

Eric Horvitz

@erichorvitz

Chief Scientific Officer, Microsoft

Seattle WA USA Katılım Mart 2011
1.6K Takip Edilen31.4K Takipçiler
Eric Horvitz
Eric Horvitz@erichorvitz·
AI models shouldn’t just agree with us, especially when we’re wrong. In clinical settings, "model sycophancy" can lead AI to mirror a clinician’s flawed reasoning, reinforcing poor decisions. As AI integration deepens, recognizing and addressing this challenge is important for delivering safe, effective decision support. Read our deep-dive study on the challenges of medical sycophancy--and mitigations: arxiv.org/pdf/2603.14158 Enjoyed the collaboration with Ivan Lopez, @BryanBunning44, @s3verett3, and colleagues. @StanfordAIMI @StanfordMed @Microsoft
English
1
2
14
1.2K
Eric Horvitz
Eric Horvitz@erichorvitz·
Exciting milestone for spatial proteomics. A multimodal AI model trained on 40M cells can generate “virtual immunofluorescence” from standard H&E slides, unlocking large-scale insights into protein activity in tumors and helping scale precision oncology and immunotherapy research. @satyanadella @hoifungpoon @jeyamariajose @naotous @MSFTResearch @Microsoft @providence @uwcse
Satya Nadella@satyanadella

We’ve trained a multimodal AI model to turn routine pathology slides into spatial proteomics, with the potential to reduce time and cost while expanding access to cancer care.

English
1
3
30
3.8K
Eric Horvitz
Eric Horvitz@erichorvitz·
An engaging audience at special AI program at @StanfordGSB, with interesting questions, curiosities, and goals around deeply understanding the implications & pace of AI influences on people and organizations.
Stanford Graduate School of Business@StanfordGSB

“I think looking back, we’ll say, ‘Wow, that’s where it all started, but we’ll still be in a time, even 20 years from now, of pretty fast-paced transformation.” @erichorvitz, Ph.D. ’91, MD.’94, Chief Scientific Officer at @Microsoft, spoke with Dean @SouleSarah as part of the GSB's applied AI initiative, AI@GSB. Read more here: stanforddaily.com/2026/02/25/mic…

English
0
0
5
1.4K
Eric Horvitz
Eric Horvitz@erichorvitz·
Enjoyed this conversation on human decision-making amidst the rising tide of highly capable AI systems, hosted by @AllDecisionEd Full panel discussion now available below. @MSFTResearch @Decision_Edu @StanfordHAI @PartnershipAI
Alliance for Decision Education@AllDecisionEd

“AI systems can help us in well-defined manner with decisions, but it's important for us to understand how and when to use AI to bolster our own decisions.” In January we were delighted to host a virtual panel, “The Human Advantage: Decision Skills in an AI-Driven Workforce.” Our Executive Director David Samuelson moderated a conversation with @AnnieDuke, @EricHorvitz, and Mike Miles on what it means to prepare students and workers for a future shaped by AI, and they kept coming back to the same idea: tools will change, but human skills, like decision skills, will endure. The recording of the webinar is available for viewing! Revisit the conversation on our blog: alliancefordecisioneducation.org/blog/watch-the… Read the “Decision Skills in the Workforce” report: bit.ly/DecisionSkills… Join our Advocates network: alliancefordecisioneducation.org/decision-educa…

English
1
1
3
939
Eric Horvitz
Eric Horvitz@erichorvitz·
“Can AI systems truly be creative?” Brian Greene asked this question in July 2024. Reply: “If people don't think AI systems are showing creativity, hold onto your seats--and check back in 18 months." At the time, development was continuing at speed, with test-time reasoning just a couple of months from being released to the public. We're coming up on the 18-month milestone now. Full conversation: youtube.com/watch?v=7wznuB… @bgreene @WorldSciFest
YouTube video
YouTube
English
0
0
1
1.3K

Keşfet

@BryanBunning44 @s3verett3 @StanfordAIMI @StanfordMed @Microsoft @satyanadella @hoifungpoon @jeyamariajose