Japanese firms take contrasting approaches to ChatGPT

Japanese firms take contrasting approaches to ChatGPT

BY KAZUAKI NAGATA

STAFF WRITER

Apr 4, 2023

As the buzz surrounding ChatGPT has grown in recent months, Japanese firms have begun to grapple with how to integrate it into their work — or whether to use the artificial intelligence-powered chatbot at all.

Panasonic Connect, which runs network equipment and security solution services, started providing an AI assistant tool based on ChatGPT to its employees in mid-February, as the company had been paying close attention to the chatbot's potential from an early stage.

“In the words of (the ChatGPT project leader at the company), it’s not about whether to use it or not, it’s about when to use it,” a Panasonic Connect spokesman said.

According to the spokesman, who declined to be named due to company policy, employees have used the ChatGPT AI assistant for tasks such as creating drafts of documents and reports, checking programming code and seeking advice on business models.

Yet given that the world is still at the dawn of the new era of AI, companies are taking different approaches on using these new generative AI tools — such as ChatGPT, Google's Bard chatbot and art generator Midjourney — with some moving fast and others taking a more cautious approach.

The launch of ChatGPT indicates the potential of human-like AI applications — which many people had believed would be tough to develop — so “the story has abruptly changed in the past several months. It’s kind of natural that a lot of people don’t really know what to do with it,” said Tadaaki Mataga, an analyst watching AI at Gartner.

With chatbot tools not yet having earned trust as a reliable service, a number of companies, especially financial institutions, are taking more of a 

With chatbot tools not yet having earned trust as a reliable service, a number of companies, especially financial institutions, are taking more of a “wait and see” approach. | GETTY IMAGES

Some tech experts have said these AI tools will likely become widespread in the future, so it is critical for business operators to embrace the trend from an early stage to strengthen their employees' familiarity with the emerging technology.

ChatGPT, developed by Microsoft-backed OpenAI, has been in the spotlight since its debut last November due to its advanced conversational skills. The app is capable of various things, such as finding and summarizing information, drafting documents and checking programing code.

Rattled by ChatGPT, other major tech companies have launched rival services. Google released its chatbot, called Bard, last month, while China’s Baidu has unveiled its own Ernie Bot.

ChatGPT’s capabilities have led many to believe that the technology is a game changer, possibly becoming the search engine of the future. Still, the chatbot is prone to making mistakes.

For that reason, Panasonic Connect has a rule that humans must check the output and make the final decision on whether it can be used.

For instance, when preparing documents, “ChatGPT could do research and write up drafts, but in the end, humans must check them and decide that they can be used,” the Panasonic Connect spokesman said.

Although the paid ChatGPT service offered by Microsoft for corporate users does not use entered data for its training, Panasonic Connect also warns its employees not to use confidential information with the bot.

“We know that it’s still at an early stage and comes with some risks, but we should deal with fear properly and give it a try while being very careful about security.”

The spokesman added that it is important to use it first and try various things to understand what the new technology is and how it can help business operations.

Rattled by ChatGPT, major tech companies including Baidu have launched rival services. | AFP-JIJI

Rattled by ChatGPT, major tech companies including Baidu have launched rival services. | AFP-JIJI

With the chatbot tools not yet having earned trust as a reliable service, a number of companies, especially financial institutions, are taking more of a "wait and see" approach.

According to OpenAI, the free version of ChatGPT uses information input by users for its training, so sensitive information could be leaked to others. But its paid version that can interface with other software applications will not use conversation inputs for training.

The three Japanese mega-banks — MUFG Bank, Mizuho Bank and Sumitomo Mitsui Banking Corp. — are taking a conservative stance, restricting their employees from accessing ChatGPT.

SMBC has a rule that employees cannot use third-party software tools including ChatGPT for their work, and MUFG has taken a similar stance.

“In terms of handling customers’ information, we don’t allow our employees to use (ChatGPT or other AI tools) for business purposes,” the bank said in an emailed response, adding that it is also researching whether such tools can be useful for their operations and how it might be possible to use them.

Mizuho Bank, meanwhile, said workers are barred from accessing third-party software services including ChatGPT from their work computers, but that it is also looking into possible future use.

In other sectors, NEC said that if employees need to input sensitive information into third-party services including ChatGPT, they need permission. But if they are not dealing with such information, they can freely use such tools. Mitsubishi Electric said its workers can access ChatGPT but are not allowed to input confidential information.

Mataga of Gartner said it is reasonable for companies to react differently at this stage.

Trust is the foundation of financial institutions’ business, so banks want to avoid critical incidents, such as leaks of customers’ sensitive information and providing erroneous information, he said.

“Because there has been a rapid flood of hype around ChatGPT, many are temporarily taking precautions, as they don’t know what's going to happen,” Mataga said.

Last week, concerns over data privacy prompted Italian authorities to temporarily ban use of ChatGPT in the country, as the chatbot service was allegedly collecting data illegally. The data protection authority said it has launched an investigation.

Kunihiro Ota, executive vice president of the University of Tokyo, released a statement Monday on how Japan’s top college should deal with generative AI tools, saying the spread of such applications would greatly impact not only educational and research activities but also the structures of industries and society.

“However, labeling generative AI as harmful and just banning its use will not solve the problems,” Ota said. “Humanity has already crossed the Rubicon in the past several months.”

As such, people at the University of Tokyo should make efforts to get ahead of the curve, find good use cases and think about how social and economic systems will change, he said, adding that the college will launch a working group to discuss such matters.

When it comes to adopting new technologies, Japanese companies, the public sector and schools tend to be conservative and eventually fall behind.

Mataga stressed that they should not be too afraid, and should balance rules and practical use from an early stage — otherwise, Japan will become a laggard again.

It’s quite similar to driving a car, he added.

“If you let people drive without a driver’s license, it’s dangerous because they would cause accidents, so it’s natural to restrict them from driving. But if they are restricted forever that would be a waste, since it would be very useful if they can master driving.”