Can the Future Astroboy Surpass the Human Child? 1. Intelligence needs a body - About Child Science

CHILD RESEARCH NET

HOME

TOP > About Child Science > Dialogues with and contributions from Specialists - Exploring new area in Child Science (1993-2007) > Can the Future Astroboy Surpass the Human Child? 1. Intelligence needs a body

About Child Science

Can the Future Astroboy Surpass the Human Child? 1. Intelligence needs a body

1. Intelligence needs a body

TAJIKA: Many countries around the world, not only Japan but also the United States, are conducting research in robotics. In Japan, this research tends to focus on the creation of humanoid robots. A typical example is Honda's ASIMO.

My book, "The Future Astroboy," explores the scientific and philosophical implications of this approach. The key scientists involved in the forefront of research in robotics are of my age, that is, in the late 40's. When I asked them why they pursued this approach, they often pointed out the influence they had from the cartoon "The Astroboy". They belong to the generation that read this manga in real time and ever since cherished the dream of creating a humanoid robot like the Astroboy.

For Japanese researchers, this quest leads inevitably to exploring what humans are through robots, and that is also what fascinates me most. As we make robots, we will actually be exploring the mechanism of emergence of human intelligence and emotion, including, possibly, mind and consciousness. In short, we will be studying humans through robots.

In the course of interviewing people for my writing, it has become clear to me why we are seeking robots that have human morphologies. The reason that robots must be modeled after humans and not like machines such as robot arms or functional rovers for planetary exploration is linked closely with the limit of artificial intelligence.

Though the computer is good at calculation, it is not at all conscious of what it is actually doing. A self-conscious computer has not been invented so far. I suppose this has led researchers to the fundamental question: can we create intelligence as software at all?

Now, we come to realize that human understanding is based on the body. For example, it would be impossible to understand what coffee is without having tasted it or just by looking it up in a dictionary. Reaching out for a coffee cup, bringing it up to the mouth, having a sip, and smelling and tasting it. In performing these actions, one perceives coffee through the body, even if one may not be able to express clearly in language. I think this is the origin of human understanding.

Michael Polanyi, a philosopher of science, has defined "intelligence" ensuing from physical perceptions as "tacit knowing". This is also related to the question of why we seek humanoid robots. It seems that tacit knowing ensues from human-like morphologies to generate human-like intelligence in a robot. In short: a robot needs a body like a human. Of course, this is just a hypothesis.


KOBAYASHI: As far as my own learning on the history of neurological research goes, it was J. Z. Young in the 1970s that had an idea to interpret functions of the brain in terms of system and information theories. What he did was synonymous with understanding humans through robots. In his book, "Programs of the Brain", Young maintains that there are a variety of programs in the human brain.

Consciousness is a difficult subject that makes up a branch of study in philosophy, psychology or psychiatry. The mechanism of consciousness, as far as my understanding goes, is something that tells us that a program is in operation. In other words, to create human-like intelligence, all we need would be a system to computerize the information that the program of artificial intelligence is in operation.


TAJIKA: That is what some scholars actually maintain. Some scientists led by Minsky, the American authority in artificial intelligence, contend that intelligence can be realized as a kind of software or program. They say, artificial intelligence can be created without a body.

But some researchers say that is not possible because intelligence needs a body. That is what Rodney Brooks of the Massachusetts Institute of Technology maintains, and Japanese researchers in robotics have been strongly influenced by his contention. Rather than trying to create a program of artificial intelligence, researchers in Japan tend to take the view that genuine intelligence cannot be generated without interaction between a program-like software and a body.


KOBAYASHI: Is that so? Well, in a study group I belong to, it was said that when visual information enters the eyes, it splits at a certain point in the brain into information destined for the visual region on one hand, and information destined for the hippocampus on the other. The information that enters the hippocampus is processed first, and then waits for the information that entered the visual region. In short, the former anticipates the latter. Such a system exists in the human brain.

TAJIKA: I see.

KOBAYASHI: Even a baby is provided with this system, I have been told. For instance, you hold out a red ball in front of a newborn and let him or her grasp it. When you repeat this movement several times, the baby ends up reaching his or her hand out to grasp the ball. When you then hold the ball just short of his or her reach, the baby reaches out again to the previous position.

TAJIKA: So the baby is anticipating the movement.

KOBAYASHI: Yes. There is a system in the brain that operates in this way. Once we clarify how this system is structured, this knowledge can be reflected in artificial intelligence and applied to robotics, step by step. However, whether we can ultimately create the human brain is a difficult question.

TAJIKA: Network structuring of the brain is a very variable operation. The basic program does not rule everything, and the network is partly formed by on-going physical actions.

KOBAYASHI: That's true. Combining information, or removing whatever is no longer needed. For example, it is said that Japanese cannot comprehend the difference between the pronunciation of "r" and "l", although babies can. This capability, however, disappears in the course of time because the Japanese language does not require the differentiation of the "r" and "l" sound. Needless to say, a new combination creates a new function. In that sense, it is natural to focus on the human body.


Profile:
Nobukazu TAJIKA
Born in Toyama Prefecture in 1953.
Freelance journalist and non-fiction writer.
Graduated from the School of Engineering, Tohoku University, majoring in architecture. Certified first-grade architect. After a career as a journalist for ""The Nikkei Architecture"", he turned to writing, publishing a series of original reports on contemporary science and philosophy from a journalistic perspective. His book ""The Future Astroboy"" (published by ASCII), the result of a two-year interview with researchers in robotics, brain science and artificial intelligence, explores a new concept of humans.

Noboru KOBAYASHI, M.D.
Born in Tokyo in 1927. Doctor of Medicine, Faculty of Medicine, The University of Tokyo
Pediatrician
Director, Child Research Net (CRN)
Director, Children's Rainbow center (Japan Information & Training Center for problems related to Child Abuse and Adolescent's Turmoil)
Professor Emeritus, The University of Tokyo
President Emeritus, National Children's Hospital

PAGE TOP