AIAS 2025: Editor's Preface

This volume contains the papers presented at AIAS 2025: AI for Accelerated Research Symposium, held on October 27–28, 2025, in San Francisco, California.

AIAS 2025 was organized to bring together researchers, practitioners, and innovators working at the intersection of artificial intelligence and scientific discovery. The symposium focused on how advances in AI can accelerate research across disciplines by improving the speed, scale, and quality of scientific inquiry. The program reflected a growing recognition that AI is no longer only a tool for analysis, but an active partner in hypothesis generation, modeling, experimentation, synthesis, and the broader research lifecycle.

The 2025 edition of AIAS placed particular emphasis on emerging directions such as foundational models for science, long-term memory mechanisms, synthetic data generation, and the automation of parts of the research process. At the same time, the symposium aimed to foster dialogue across fields, connecting AI researchers with experts from the physical sciences, life sciences, medicine, engineering, and related domains. The result was a program intended not only to showcase technical advances, but also to strengthen a broader community committed to accelerating science responsibly and effectively through AI.

AIAS 2025 received 55 submissions. Each submission was reviewed by members of the Program Committee, with an average of 2.2 reviews per paper. After careful evaluation, the committee selected 24 papers for presentation and inclusion in these proceedings.

As a symposium, AIAS is meant to serve both as a venue for presenting high-quality research and as a forum for shaping an emerging field. We hope these proceedings capture not only the technical contributions of this year’s authors, but also the energy of a community working to define how AI can meaningfully contribute to scientific progress.

We are deeply grateful to the authors for submitting their work and to the Program Committee and reviewers for their time, rigor, and thoughtful feedback. Their efforts were essential to the quality of the symposium and these proceedings. We also thank the organizing committee, speakers, session organizers, and all participants who helped make AIAS 2025 a successful and stimulating event.

We would like to extend special thanks to the Tianqiao & Chrissy Chen institute to organize the symposium. The symposium is partnered with UC Berkeley College of Computing Data Science, and Society. The partnership helped make this event possible and contributed to building a forum for interdisciplinary exchange on AI for accelerated research.

We also appreciate the support of EasyChair, which assisted in the management of the submission and review process for the symposium.

We hope readers find these proceedings valuable, thought-provoking, and representative of an exciting and rapidly developing area of research.