[ITmedia Executive] TEPCO's AI reform that started because "there was no one to use it" - The CIO talks about how to think about business processes, role division, and investment recovery

The Tokyo Electric Power Company Group, one of Japan's largest infrastructure companies, was one of the first to promote an "AI shift" throughout the company after the advent of generative AI. However, even after introducing the tool, it did not spread widely in the field. How are you trying to overcome that wall?

In the keynote speech at Sansan Innovation Summit 2026 held by Sansan on October 14, 2026, Mr. Michi Seki of Tokyo Electric Power Company Holdings (Managing Executive Officer, Chief Information Officer [CIO] and Chief Information Security Officer [CISO]) spoke. The interviewer was Chikahiro Terada (President/CEO/CPO) of Sansan.

Mr. Seki describes the emergence of generative AI as, ``Some people say it's a revolution on the same level as the Internet, but I say it's a revolution on the same level as the invention of electricity 150 years ago.'' The view is that the transition from fire to electricity is now occurring again. That's why I have told everyone within the company that I want them to make their own changes, rather than just making systems more efficient and improving business operations.

Even so, the number of monthly active users (MAU) immediately after its introduction was sluggish. When we asked the reasons for this in an internal survey, the answers we received were divided into four categories. I don't know how to use it. No good results. I'm afraid to include personal information or company regulations. And there's no one to use it.

How to use it can be trained, and the quality of results can be improved by learning internal data. It's only natural for electric power companies to be careful about how they handle information, and to respond to people saying it's scary, they can respond by putting in place rules. It was the last voice that troubled Mr. Seki.

"The problem was the fourth one, and the opinion was that there was no one who would use it." (Mr. Seki)

Why is there no one using it? Mr. Seki believes that the starting point is that current business processes do not have a step to use generative AI in the first place.

Therefore, TEPCO has set a goal of calling a business process based on the use of generative AI its "North Star." However, this cannot be achieved in one step. The idea is for each and every employee to start with small efficiency improvements and gradually advance sophistication, ultimately creating a world where people and AI advance together.

An example of this is the on-site support app ``Safe Now''. The company often works at construction sites, etc., so when users take a single photo of the site with their smartphones, AI identifies the source of danger, estimates the risk, and even suggests specific measures to take, such as placing road cones. The system identifies unsecured ladders and trucks parked diagonally on slopes as sources of danger, and evaluates the severity and likelihood of the danger occurring. Mr. Seki explains that this is an example of utilizing AI's three major functions: recognition, prediction, and generation.

However, when you introduce AI, you tend to rely on them to "do something." To prevent this from happening, Mr. Seki says he is currently focusing on thinking about ``what people do.'' In the next chapter, we will look at Mr. Seki's thoughts on the division of roles.

So how should the roles be divided between humans and AI? Mr. Seki declined, calling it ``a completely personal opinion,'' and said the following.

"Generative AI basically only knows the 'past.' I think that a world where humans and AI are fused requires thinking about the division of roles." (Mr. Seki)

AI can perform searches, summaries, and routine tasks because they can be derived from past data. Mr. Seki believes that in the future, AI will be able to do everything except for things that only humans can do. That is why I believe that humans should take on the task of envisioning the future, while being conscious of areas where AI cannot outperform humans.

There have been similar cases in the past. Vacuum cleaners and washing machines eased the burden of household labor and supported women's advancement into society. The same goes for the Internet. Mr. Seki hopes that generative AI will become a tool for ``speed to innovation,'' which shortens the time it takes for people to innovate.

On the other hand, we are taking a calm view of the company's management. "AI is not free. It costs money." With so many management issues piling up, it is often dismissed as ``generative AI will solve all of them,'' but Mr. Seki's position is to consider the essence of ``cost effectiveness.'' When you receive a report from your subordinate that ``the number of inquiries has been cut in half,'' ask them back, ``Where did that effect go?'' The question is, are you wasting your free time adjusting the appearance of PowerPoint?

Therefore, I am currently choosing projects that clearly show up on the profit and loss statement (PL). For example, it is a project that can cut commission costs in half. Then, check whether it really works, whether there is data, whether there are tools, and whether there is someone who can lead the way.

I want to recover my investment as quickly as possible. With this in mind, Mr. Seki is constantly reducing costs and creating value so that cash flows within his budget.

On the other hand, Seki is optimistic about the AI ​​threat theory. "I'm not really on the risk side, and I think humanity will outgrow this." When electricity first appeared, it was said to be dangerous, but now safety regulations have been put in place and everyone can use it safely. By teaching ethics to generative AI and increasing supervision, he says, ``Humanity can turn it into our 'friend.''

TEPCO's AI shift, which began with the complaint that "there is no one to use it," is supported by three perspectives: reorganization of business processes, division of roles between people and AI, and investment recovery.

Finally, when asked to give a message to the participants who are taking on the challenge of AI shift, Mr. Seki began by saying, ``Whether it's DX or AX, I believe that all of you are innovators.'' Innovators must be strong individuals, and there are three elements necessary for this.

The first thing to do is to correctly recognize the facts of your company's situation and make decisions calmly. It refers to the attitude of accurately looking at the current situation without being influenced by assumptions or atmosphere. Next is a sense of purpose. Mr. Seki asks, ``What are you doing this for?Are you doing it for the sake of the customer?Are you doing it to install tools?'' Mr. Seki asks. If the introduction becomes an end in itself, there is a risk that we will return to the situation described at the beginning of the article, ``no one will use it.''

The last is what Seki calls "art thinking." ``Can you suddenly draw a picture on a white canvas?'' Ask yourself what kind of image you want to create and whether you have the passion to make it a reality.

"Change is quite lonely and painful. If you have passion, you can keep going." (Mr. Seki)

Passion makes you wake up early in the morning, and purpose helps you sleep well at night. Mr. Seki said this, and appealed to the audience, ``I hope you will work towards change with passion and purpose.''