TOKYO: Japanese digital minister KONO Taro is eager to promote the active use of generative artificial intelligence by central government staff for work, expecting “a big benefit” from it.
The use of generative AI will provide “a great benefit at central government workplaces as long as learning data are handled carefully,” Kono said in a recent interview with Jiji Press.
Kono said office work would be done far more easily by utilizing generative AI, which automatically creates texts and images. For example, by inputting the full text of a government white paper into an AI engine, officials would be able to easily create a summary or PowerPoint presentation slides, Kono said.
“We need to give government staff more time to do creative work. It may even be possible to have AI do more and more of the less creative but necessary (routine) tasks,” he said.
Kono said that generative AI may be able to make drafts of parliamentary statements to be read out by cabinet ministers though they would likely be very moderate and boring.
“The most important thing is to decide how deep such statements should go,” Kono said.
Parliamentary debates should be invigorated if government staff can spend more time considering whether and how much ministers should say beyond what are written in drafts of such statements, he stated.
“That’s what the government really needs to do,” Kono stressed.
Asked about demerits of AI, Kono said: “We must handle data properly. It’s necessary to consider the safety of data and the protection of personal information including how to handle sensitive data.”
“But merits will far outweigh demerits at least for central government staff as long as data are handled carefully,” he added.
In May, leaders of the Group of Seven advanced economies agreed in their summit in the western city of Hiroshima to start discussions to create international rules on generative AI tools under the Hiroshima AI Process framework.
“There is a debate over how Japan should use platforms and whether the country should develop them on its own or with other countries,” Kono said.
It should be better to have Japanese-made AI, he said, adding that promoting domestic development will help the AI market expand and technologies improve.
On regulation of training data for generative AI, Kono said Europe believes the use of such data should be restricted out of human rights and privacy concerns, while Japan, the United States and Britain think that restrictions should be set only when needed as they prioritize the development of AI.
Regarding the issue of generative AI spreading misinformation, Kono said, “If the spreaders try to break regulations, we’ll need to fight back using technologies.”
Asked whether Kono himself has ever used a generative AI tool, he said he utilized ChatGPT to write a congratulatory speech for the conferment of a former member of the House of Councillors, the upper chamber of the country’s parliament.
The AI chatbot produced a very good speech text that mentioned the former lawmaker’s admirable personality and focus of activities, Kono said.