Multimodal AI and New Collaborations at AIMS26
June 23–25, 2026 · UC San Diego, La Jolla, California
I joined AIMS26: AI for Multimodal Science, a three-day workshop that brought together researchers from different institutions and disciplines to explore how multimodal AI can support scientific discovery.
The program combined invited talks, coding tutorials, lightning presentations, a collaborative hackathon, and professional-development sessions. The talks and discussions moved across biology, medicine, materials science, physics, climate science, and computer vision. What made the workshop especially useful was the chance to move between these fields while returning to shared questions: how to integrate different kinds of data, how to design models that work across modalities, and how to turn those methods into scientific insight.
The hackathon made the technical ideas more concrete through a hands-on challenge involving 3D reconstruction and the fusion of images, radar, and LiDAR data. Sessions on grant and fellowship writing and scientific entrepreneurship added another useful dimension, prompting conversations about how research ideas can be communicated, supported, and developed beyond the lab.
The conversations between sessions were just as valuable. I came away with new technical perspectives, a better understanding of how researchers in other fields approach AI-enabled discovery, and connections that I hope will grow into future collaborations and lasting friendships.
I am grateful to the AIMS26 organizers, speakers, and contributors—including Yasmin Kassim, Konstantinos D. Polyzos, Xiaoyu Zhao, Eleonora Rachtman, Tara Javidi, Uri Manor, and many others—for creating such a thoughtful and welcoming program. My thanks also go to Schmidt Sciences for supporting the workshop and to the wider UC San Diego community for hosting us.