Lab and industry leaders: open and proprietary AI models will coexist for science
Summary
Panelists argued that both open models and proprietary frontier models have roles in scientific research, with open models powering broad token consumption and proprietary models advancing frontier reasoning for specific high‑investment tasks.
Panelists debated the roles of open‑source and proprietary AI models in scientific research, concluding that both will be needed for different workloads.
Misha Laskin, cofounder and CEO of Reflection, said open American models are essential so labs and scientists can customize systems toward lab‑held data. "If you have great open models, then scientists will be able to customize it for their needs," he said. John Serrao said national labs use a "managed research environment" rubric to balance openness and protection where national‑security or sensitive data is involved.
Kartik Narayan said frontier proprietary models require large investments and remain necessary for high‑reasoning tasks, while open models will likely dominate token consumption because of cost effectiveness. The panel recommended a heterogeneous platform approach that can run open weights and proprietary inference depending on task requirements.
AI generated
The text on this page is AI generated. Summaries, highlights, analysis, and video transcripts are all produced from the original source material.
AI can make mistakes, so if you spot one, and we will fix it for everyone.
Note: the source content is unaltered by us. Any content source we link to, be it a video, an audio recording, or a document, is presented exactly as its publisher released it. That publisher is usually a government body, sometimes an individual official or another organisation.
