Research at AIIT

Dialog with an Uncertain Future / Excerp (Irregular updates)

The 13th Installment“The Ability to Recognize Successful Technology When You See It”

by Shigeru Shimada,
Professor, Master Program of Information Systems Architecture

I hope you are not working under the mistaken impression that you will definitely succeed if you can develop technology that no one has come up with before. You may succeed in developing a particular technology, but it won't be a true success unless society accepts it; it will only be discarded sooner or later as useless technology. So how do we recognize successful technology when we see it? This article will provide examples while discussing how technological developments should be undertaken in the field of IT.

Before coming to this school, I worked at a corporate research lab undertaking technological developments for close to thirty years in the field of AI (Artificial Intelligence). Looking back, the research lab where I worked allowed a level of freedom in its activities that was more like that of a university than a corporate organization, and it was a great environment in which to develop technology. The mission of the lab was, “Do not rely on the technology that is immediately available; create trends with an eye on the next ten years.” Under that command, we were allowed to pursue every conceivable possibility in our own style of development in everything we did, from analyzing current technology and demands to identifying issues, developing technology, and assessing compatibility with demand. Even in an environment of such freedom, technological developments do not succeed easily. Even if you manage to develop a particular technology, it will not be recognized as successful from a corporate perspective if society does not accept it (in other words, if it does not sell).

Let us take a paradoxical view and consider some useless technologies. While a typical example would be an airplane that flaps its wings, let me give you an example of a failure in the field of IT that I experienced during my days at the lab. Thirty years ago, handwriting recognition was considered to be at the forefront of AI research, and a global competition was underway to increase the level of pattern recognition technology that measures its success. Our group believed that the mere recognition of text would have only limited application, and we worked on the challenge of recognizing design drawings and maps that involved geometrical recognition. One day, a client consulted us regarding his desire to print out what was written on a blackboard, so we concentrated on developing a technology that would use geometrical recognition to automatically identify shapes as vector data and words as text code. But what a person writes on a blackboard are words and shapes that all vary in terms of the angles and sizes in which they are written, and the preliminary process of separating them into words and shapes would take several minutes even using a supercomputer, which was far from practical. While we were struggling with these issues, one Japanese vendor commercialized the so-called “whiteboard system,” which attached a scanner above a sheet of rolled white paper and printed out a reduced image of words and shapes written as they appeared on the board. This system merely printed handwritten words and shapes as they appeared on the whiteboard onto to a document-sized printout, but consumers embraced it, since the printout only took about ten seconds while the device scanned the whiteboard. The patent for this system eventually went on to win the Japan Patent Award and record explosive sales nationwide, and I heard that numerous orders were received from overseas clients. In other words, it is a typical example of successful technology.

Our group was initially aware of the image scanning and reducing technologies, and at one point had considered filing a patent for it, but the patent team in our company had refused to file the application, considering it too obvious. I still look back with regret and wonder why we restricted ourselves to developing such useless technology, and my understanding now is that we lacked sufficient understanding of consumer needs and had underestimated the human ability to recognize patterns intellectually. The important point in terms of consumer need was to print out whatever was written on a huge blackboard as a document, and in terms of understanding what was written, humans could do it much more efficiently. It seems that we were unable to take either perspective.

Applying this discussion to the impracticality of an airplane with flapping wings, we realize that too much emphasis on attempting to make the flying mechanism emulate biological action tips the balance between the essential function of flying and the cost of making it happen. Although it has been announced that the latest military weapon development has achieved an agile flying capability in the form of a plane with flapping wings, which prevents us from rejecting the idea entirely, we must note that even so, the need is different from that of the former.

I have heard that intellectual technology, as seen in the next-generation Web 3.0, will become the core technology of the future. I feel that the skill of recognizing successful technology when you see it –balancing technology with human capabilities rather than attempting to automate everything – will become more important than ever in the field of IT.

I hope that this discussion about my mistakes has provided you with a lesson in the secrets of successful technology.

Back to index page

  • The 14th Installment

    “The Birth of Industrial Design”

    by Yoshie Kunisawa,
    Professor, Master Program of Innovation for Design and Engineering
  • The 13th Installment

    “The Ability to Recognize Successful Technology When You See It”

    by Shigeru Shimada,
    Professor, Master Program of Information Systems Architecture
  • The 12th Installment

    “Idle Talk”

    by Yoshinori Kanno,
    Professor, Master Program of Innovation for Design and Engineering