Dr. Kazuo Yano Recommends “Endlessly Experimenting”

At an AIWS Round Table in Japan earlier this month, top Hitachi engineer Dr. Kazuo Yano gave the keynote speech and called for a switch from standardization to experimentation. Dr. Yano is Chief of Hitachi’s R&D Group and has worked with AI systems for nearly 15 years.

As part of his speech, he showed an AI system learning to swing, much as a human would on a swing-set. While it starts off with haphazard motions, it quickly learns to swing faster and higher faster than a human could. AI “keep experimenting,” said Dr. Yano, while humans stop after a certain point. “That’s our problem,” he said, “we must endlessly experiment and learn.” Whereas many models currently favor standardization and rigid rules, he believes that what’s needed now is diversity and testing.

Instead of replacing labor, AI really threatens to replace rules, believes Dr. Yano. The key to successful AI policy and paradigms is an “outcome-oriented” approach instead of a “rule-driven” one.

Watch Dr. Yano’s keynote speech below: