McKinsey just dropped its 2025 AI report. 1. Everyone’s testing, few are scaling. 88% of companies now use AI somewhere. Only 33% have scaled it beyond pilots. 2. The profit gap is huge. Just 6% see real EBIT impact. Most are still stuck in “experiments,” not execution. 3. The

1/ AI Investment Boom 🤖 Big Tech is embarking on an epic AI data center buildout, with capital expenditures of ~$400 billion this year and trillions more on the way.

Strong kick-off @StanfordAIMI of the Inaugural AIMI AcademicxIndustry Summit 2025 on a nice fall morning @curtlanglotz #StanfordFacultyClub @StanfordAIMI

Google introduces Test-Time Diffusion Deep Researcher Don't sleep on diffusion models. Test-Time Diffusion Deep Researcher (TTD-DR) is a deep research agent that models research writing as a diffusion process. Instead of static reasoning or bolted-on tools, the system drafts

What an extraordinary day & a half at the AIMI Symposium & inaugural Pediatric Symposium! Thank you to all speakers & participants for pivotal discussions. We look forward to seeing you at our future meetings! #AIMI25 #StanfordHealthAIWeek

🚀 We hosted an exclusive panel discussion on AI in Healthcare, featuring leading voices at the intersection of technology and life sciences! Moderated by @shivaamiri, PhD, Partner and VP – Head of AI and Data Intelligence at Pivotal Life Sciences, the panel brought together:

Agency > Intelligence I had this intuitively wrong for decades, I think due to a pervasive cultural veneration of intelligence, various entertainment/media, obsession with IQ etc. Agency is significantly more powerful and significantly more scarce. Are you hiring for agency? Are

Learn about AI foundational models in radiology in this review article! @StanfordAIMI @StanfordRad @StanfordMed @magdapasc @zhjohnchan @mayavarma23 @loublanks @Alaa_Youssef92 @cxbln @curtlanglotz @Dr_ASChaudhari https://t.co/O9BQ5BQxvw

When building #ML/#genAI models, we often assume that expert-labeled data is consistent and reliable. But that’s not always the case. In radiology, for example, data labeling has a subjective component—annotations often reflect the clinician's perception rather than an objective

Details of the first federated learning model deployed by our friends at @NASA 🛰️ between Earth and the ISS is now out! Some really cool work by @RyanTScott and team!

Amazing turnout at the AIMI Fall Open House! Grateful to connect with & support Stanford's growing AI in health community. Highlights included opportunities to engage, vibrant discussions & a shared vision for advancing the field. Looking forward to our next community event!

We're excited to share a new paper on the AIMI Center's journey over the past 6+ years, highlighting how we built an interdisciplinary hub for AI in medicine & the core pillars guiding our work. Read more: https://t.co/2lCgc5nRWF #AIinHealthcare #StanfordAIMI

Another update to the HELM benchmarks (MMLU, Lite, AIR-Bench) with new model versions. We do see some movement near the top even with these "minor" versions updates. https://t.co/kJ9AXKzGF3 Claude 3.5 Sonnet (20241022) Gemini 1.5 Pro (002) Gemini 1.5 Flash (002) GPT-3.5 Turbo

Sanjeev Sockalingam giving a plenary at #aadprt2020 on "Avoiding the Curriculum Carousel: Approaches to Curriculum" @uicdme @theWilsonCentre

Our latest work improves chest x-ray report generation with LLM alignment tools! We build a scalable preference fine-tuning pipeline to improve automated metrics + expert rad. win rates—all WITHOUT any radiologist feedback. Preference alignment w/o needing human preference. 1/3

If I finetune my LM just on responses, without conditioning on instructions, what happens when I test it with an instruction? Or if I finetune my LM just to generate poems from poem titles? Either way, the LM will roughly follow new instructions! Paper: https://t.co/Jk3EOtLJXF

This was a really fun project. Fine-tuning a model on "" => response produces a model that can do instruction => response What??

Big news! Our co-director @drnigam & teams published a groundbreaking paper on Evaluating Fair, Useful, and Reliable AI Models (FURMs) in Health Care. These open-source tools redefine AI evaluation in healthcare. Amazing work! More here: https://t.co/cXdBsuAPvj #AIinHealthcare

#MentalHealth as the top health concern around the globe. Compelling data from recent survey..."Ipsos interviewed a total of 23,667 adults aged in India, Canada, Ireland, Malaysia, South Africa, Türkiye, US, Thailand, Indonesia, Singapore..." https://t.co/PvW3V9tqWE

📢 Introducing HeadCT-ONE: Our new paper addresses a major gap in AI evaluation for radiology—capturing semantic equivalence. Using ontologies, we standardize medical terms, making AI-generated head CT reports more accurately comparable, even when phrasing differs.🧠✨