Issei Yamamoto became one of Japan’s best-known developers of artificial intelligence when his algorithm defeated the top-ranked player of Japanese chess. Now, he’s pursuing an even more challenging task of human emulation: achieving a fully self-driving automotive system.
“There’s no reason we can’t do it,” Yamamoto, chief executive officer of Turing, said in an interview in Tokyo. “I keep telling people, ‘We can do it too.’” Turing plans to unveil a self-driving car with at least 30 minutes’ driving range next year and develop a fully autonomous car by 2030. To achieve full automation, Turing’s engineers are adopting the more ambitious approach of training their AI to learn everything by itself, shying away from rule-based algorithms, according to Yamamoto. While rule-based systems are simpler to develop, they have limited capability when handling complex tasks and uncommon scenarios. Heron’s machine learning program will pursue sophistication close to human intelligence, Yamamoto said.