When Japanese business mogul Masayoshi Son, who founded SoftBank Group, took the stage at his firm’s event in October, he spoke passionately about the boom in generative artificial intelligence, and asked a question:
“Please raise your hand if you use ChatGPT almost every day for work?”
Seeing that the people who did was less than 10% of the audience, Son castigated the remainder: “This is bad! If you didn’t raise your hand, you should be repentant and rethink your life.
“Whether you like it or not, the AI revolution will come,” Son said, adding that he has been heavily using the chatbot developed by OpenAI.
Generative AI stole the show in the tech sector in 2023, with Son and many business leaders touting it as the next big thing, urging people to get onboard with the technology to avoid being left behind.
In Japan, both the public and private sectors have been somewhat slow to make progress on using digital technology to boost productivity and transform business models, but the rise of generative AI may be impetus for them to play catch up.
In a report about Japan’s strategy for its digital industries released in June, the Ministry of Economy, Trade and Industry said that — amid the rise of generative AI, quantum computing and other game-changing technologies — now is the “last chance” for the nation’s industries to realize a digital transformation.
“Being left out from this trend would be an existential matter,” the report said.
In recent years, "digital transformation" has been a buzzword, with the use of AI being one of the key aspects.
“But in my view, the progress has been slow in terms of effectively integrating it into daily operations,” said Shimpei Miyoshi, partner at PwC Consulting, who is well-versed in data analytics and AI.
One of the reasons for the limited steps forward has been Japan's conservative corporate culture, which is risk-averse and engenders caution about new tools such as AI, Miyoshi pointed out.
Before generative AI rose to prominence, AI was only used by a small pool of data scientists and engineers, since its processes and logic were too complicated for people who weren't tech savvy to understand, he said. For instance, even if companies produced sales projections or marketing strategies using AI, decision-makers were doubtful of the results because it was difficult to explain how such conclusions had been reached.
As for generative AI, many firms initially grappled with how to integrate it into their work — or whether to use the chatbot at all — as they were concerned that the new tool might entail some serious risks, including leakage of sensitive information.
Yet the situation appears to be changing.
In a survey conducted by PwC in the spring, 44% of the 238 respondents said they did not know about generative AI at all, and only 10% said they had used it. But another survey from the fall showed just 4% of the 912 respondents did not know what generative AI was, while 73% said they had used it.
What’s monumental about generative AI is that its user interface is simple, with many tools including ChatGPT being text-based chatbots that don’t require technical skills, such as programming, to use them.
Still, even a Japanese tech powerhouse like NEC was initially cautious, as it was unclear whether generative AI was really a game-changing technology, which would entail some risks.
“Our employees were wondering whether they could use generative AI for their work and there were quite a lot of discussions,” said Toshifumi Yoshizaki, chief digital officer at the company.
But NEC swiftly hammered out generative AI guidelines for employees in April and also developed a generative AI tool for in-house use the following month.
Over 20,000 workers have used the generative AI system, helping them to, for example, drastically cut the time it takes to prepare documents and put together minutes of meetings.
NEC has also used its generative AI system to beef up cybersecurity by having it analyze the results of cyberthreat examinations and create fake spam emails to train employees.
“The situation has completely changed in just about six months,” Yoshizaki said.
Based on a conviction that generative AI will be an essential part of daily business for any company, just like the internet or smartphones, NEC has developed its own enterprise-focused generative AI system called Cotomi.
Cotomi is better at processing Japanese than some major chatbots developed overseas, the firm claims.
Taking advantage of this Japanese aptitude, NEC plans to produce versions of Cotomi that are tailored toward different industries, such as manufacturing and health care, with the chatbot able to understand sector-specific terms. NEC is aiming for ¥50 billion ($355 billion) in sales from its generative AI business over three years.
Still, there is a question of how far companies will push such tools. Many firms tend to take a superficial approach to digitalization, being satisfied with simply promoting paperless or holding meetings online, rather than focusing on a deeper transformation of their business or working style.
In that sense, the accessibility of generative AI — which is used by typing in questions or commands — means that users can grow more confident as they use it and work more efficiently. That is actual digital transformation, Yoshizaki said.
PwC’s Miyoshi said that while enhancing the efficiency of existing tasks is critical, business operators should also review the necessity of them, since improving the efficiency of activities with little value won’t actually help boost a company's productivity.
What is important for companies, Miyoshi said, is to seek out the unique opportunities that only generative AI is capable of providing.
Based on conversations with clients, “we think that companies that can drastically improve work efficiency and also find ways to produce value that only generative AI can will be able to grow a lot.”
Generative AI is also driving digital transformation in the public sector, which is notorious for antiquated practices and being slow to change. For instance, the use of hanko seals and fax machines drew flak for hampering efforts to minimize bureaucracy during the pandemic.
But there are exceptions: In April, the city of Yokosuka in Kanagawa Prefecture became Japan's first municipality to run a trial of ChatGPT for its administrative operations.
“For promotional purposes, we were aiming to become the first local government to use ChatGPT,” said Kohei Ota, an official at the city’s digital government promotion office.
Yokosuka was not necessarily a laggard in terms of using technology to improve efficiency. For instance, the city is known for having digitalized various administrative services, such as the issuing of residence certificates.
The city was able to act quickly on generative AI because it was fairly easy to incorporate ChatGPT’s functionality into a chat tool that the city was already using. The city’s mayor also saw the potential of generative AI and instructed his officials to consider how ChatGPT could be useful.
The one-month trial targeting the entire city office, which employs about 3,800 officials, turned out to be quite positive. A survey answered by 433 staffers found that about 80% of them thought ChatGPT helped improve efficiency, and 76% said they would like to continue using it.
ChatGPT has become a regular work tool in Yokosuka, with the volume of processed text staying at around 20 million characters a month.
“I often hear from other municipalities that people are really keen to use generative AI tools at first, but many lose motivation after a while,” Ota said.
“To be honest, we had expected that the level of usage would decrease, so we have been focusing on keeping officials motivated by publishing newsletters and running training sessions and contests.”
Generative AI has also helped Yokosuka ease some time-consuming administrative tasks.
For example, a number of workers at the fire department are unaccustomed to clerical work, so it takes time for them to write documents that are organized and easy to understand. But they have been able to do this more quickly based on drafts ChatGPT produces.
Some city officials have even been asking ChatGPT to suggest new policy ideas, including on how to bolster local tourism, generating possible initiatives that they would not have been able to come up with otherwise.
Yokosuka is sharing its experience with ChatGPT and the resulting know-how with other municipalities, with the aim of helping generative AI spread more widely in the public sector.
“There is only so much one municipality can do with its knowledge and know-how, so we want to cooperate with others,” Ota said.
“We are hoping that this will eventually lead to promoting digital transformation at local governments nationwide.”
With your current subscription plan you can comment on stories. However, before writing your first comment, please create a display name in the Profile section of your subscriber account page.