The Micro (Transformer) and Macro (Agent) of AI for Science
— Domain-Specific Strategies at Both Levels —
April 28, 2026 (Tue) 16:00-17:00
AIRDS invited Mr. Yuma Nakamura, a member of the Aspuru-Guzik Lab at the University of Toronto known for its AI research, during his visit to Japan, and held this online lecture. Despite the short notice, the event drew far more interest than expected, with 132 registrants and 119 attendees on the day, and concluded with great success. We sincerely thank everyone who took part.
Overview
Date & Time
April 28, 2026 (Tue) 16:00-17:00
Format
Online broadcast from Room 31B, 3F, Engineering Bldg. 2, Hongo Campus, The University of Tokyo (Teams webinar)
Abstract
The keyword "AI for Science" has recently attracted significant attention, but the concept is often too broad. In this seminar, the speaker broke down this vast theme from the perspective of AI architecture into two parts — "Agent" (macro-level autonomous coordination by groups of language models) and "Transformer" (the micro-level core deep-learning technology of language models) — and discussed them using chemistry, biology, and quantum computing as concrete examples.
Speakers
Lecturer
Mr. Yuma Nakamura
Ph.D. student, Department of Computer Science, University of Toronto
Ph.D. student in the Department of Computer Science at the University of Toronto (enrolled September 2025), supervised by Prof. Alan Aspuru-Guzik. He specializes in applications of AI and quantum computing in drug discovery. Previously, he worked as a data scientist at IBM for six years, leading data analysis and machine learning projects in the healthcare domain. He earned master's degrees (in physics and chemistry, respectively) from Tohoku University and Tsinghua University (China) through a double master's program (2015-2018).
Moderators
Prof. Keisuke Nagato
Professor, The University of Tokyo; AIRDS Director
Mr. Naruki Yoshikawa
Senior Researcher, National Institute for Materials Science; AIRDS Secretary
About AIRDS Membership
The AI Robot-Driven Science Initiative (AIRDS) is a non-profit general incorporated association established to redefine the process of scientific research with AI and robots and to accelerate its progress.
Researchers, engineers, and companies from industry, academia, and government gather across disciplines to promote activities such as developing research infrastructure and international collaboration.
By joining as an individual member, you can enjoy the following benefits: