Role of Graduate Education in the Age of AI
Generative AI has become increasingly prevalent in graduate-level education and research. From summarizing vast amounts of literature to generating code, AI can now complete tasks in seconds that previously required days of effort. I am truly astonished by these capabilities, to the point where it prompts me to reconsider the fundamental role of graduate education. However, in the shadow of this overwhelming efficiency, we tend to overlook a crucial truth: while information can be copied, experience cannot. What, then, is required to transform information into true "intelligence"?
Information can be copied, experience cannot
A hint lies in the famous words that theoretical physicist Richard Feynman left on his blackboard at Caltech: “What I cannot create, I do not understand.” Simply having AI generate an answer is akin to observing a finished product created by someone else. According to Feynman’s philosophy, one cannot truly claim to understand something unless they can reconstruct the process from scratch with their own hands.
He also wrote on that same blackboard: “Know how to solve every problem that has been solved.” Even for problems already solved by others—or by AI—attempting to solve them again yourself is vital. It is within these "detours" and failures that irreplaceable experience truly resides.
The Importance of experience
This mindset of "reconstruction" is equally essential when interacting with generative AI. Rather than simply copy-pasting AI's output, we must probe why it reached a certain conclusion, experiment under different conditions, and rebuild the logic ourselves. This gritty, iterative process of dialogue is the very essence of the "experience" required of a researcher.
If we allow ourselves to be swept away by the wave of efficiency and rely solely on shortcuts, we risk eventually losing the "ability to formulate our own questions"—a treasure and a joy that is essential to researchers and, ultimately, to human beings. While harnessing the powerful capabilities of AI, we must never forget the sensation of walking the ground, one step at a time. True horizons of new knowledge can only be glimpsed by those who have accumulated such irreplaceable experiences. This is a pursuit that can arguably only be fully realized within the abundant time and environment afforded by graduate school. Furthermore, dialogue with fellow students and faculty is of the utmost importance. The intellectual excitement and the expansion of curiosity derived from truly "understanding" are experiences unique to human beings.
Strength of our graduate school
Our Graduate School conducts world-class research and education across four key areas—Data Science, Computational Science, Information Security, and Health and Medical Sciences—as well as their multidisciplinary intersections. We offer an environment that is exceptionally rare for a public university, providing students with easy access to high-performance computing resources, including the RIKEN supercomputer "Fugaku" located adjacent to our campus. While the number of graduate schools specializing in "Information Science" or "Data Science" has increased rapidly in recent years, our human resources and research environment remain distinct features of this Graduate School. For further details, please refer to our website or brochure. We look forward to welcoming many prospective students and fostering new collaborative research projects.
Generative AI has become increasingly prevalent in graduate-level education and research. From summarizing vast amounts of literature to generating code, AI can now complete tasks in seconds that previously required days of effort. I am truly astonished by these capabilities, to the point where it prompts me to reconsider the fundamental role of graduate education. However, in the shadow of this overwhelming efficiency, we tend to overlook a crucial truth: while information can be copied, experience cannot. What, then, is required to transform information into true "intelligence"?
Information can be copied, experience cannot
A hint lies in the famous words that theoretical physicist Richard Feynman left on his blackboard at Caltech: “What I cannot create, I do not understand.” Simply having AI generate an answer is akin to observing a finished product created by someone else. According to Feynman’s philosophy, one cannot truly claim to understand something unless they can reconstruct the process from scratch with their own hands.
He also wrote on that same blackboard: “Know how to solve every problem that has been solved.” Even for problems already solved by others—or by AI—attempting to solve them again yourself is vital. It is within these "detours" and failures that irreplaceable experience truly resides.
The Importance of experience
This mindset of "reconstruction" is equally essential when interacting with generative AI. Rather than simply copy-pasting AI's output, we must probe why it reached a certain conclusion, experiment under different conditions, and rebuild the logic ourselves. This gritty, iterative process of dialogue is the very essence of the "experience" required of a researcher.
If we allow ourselves to be swept away by the wave of efficiency and rely solely on shortcuts, we risk eventually losing the "ability to formulate our own questions"—a treasure and a joy that is essential to researchers and, ultimately, to human beings. While harnessing the powerful capabilities of AI, we must never forget the sensation of walking the ground, one step at a time. True horizons of new knowledge can only be glimpsed by those who have accumulated such irreplaceable experiences. This is a pursuit that can arguably only be fully realized within the abundant time and environment afforded by graduate school. Furthermore, dialogue with fellow students and faculty is of the utmost importance. The intellectual excitement and the expansion of curiosity derived from truly "understanding" are experiences unique to human beings.
Strength of our graduate school
Our Graduate School conducts world-class research and education across four key areas—Data Science, Computational Science, Information Security, and Health and Medical Sciences—as well as their multidisciplinary intersections. We offer an environment that is exceptionally rare for a public university, providing students with easy access to high-performance computing resources, including the RIKEN supercomputer "Fugaku" located adjacent to our campus. While the number of graduate schools specializing in "Information Science" or "Data Science" has increased rapidly in recent years, our human resources and research environment remain distinct features of this Graduate School. For further details, please refer to our website or brochure. We look forward to welcoming many prospective students and fostering new collaborative research projects.
Yoshi FUJIWARA
Graduate School of Information Science, Dean, Professor, Ph.D.
