Partnering with generative AI to co-create a happier society. The challenge of Hitachi, Ltd. and the Dentsu Group.

It's Not Just About Boosting Productivity! What Is the True Potential of Generative AI?

──Please tell us about the "AI for EVERY" project overview and what prompted its launch.

Yoshida: "AI for EVERY" is a project we've been advancing as a three-way collaboration since around autumn 2024. We believe that by combining the strengths of the Dentsu Group and Hitachi, Ltd. (hereafter Hitachi), which have different areas of expertise, we can provide generative AI services that solve various societal challenges.

For over 100 years since its founding, Hitachi has made solving societal challenges its mission. Therefore, regarding generative AI, we are pursuing applications that more effectively address these challenges. Currently, Hitachi utilizes AI across a wide range of fields, including demand forecasting and fault prediction in areas such as finance, manufacturing, distribution, and infrastructure.

──The current approach to AI utilization is mainly focused on replacing tasks that humans can do, thereby improving operational efficiency, right?

Yoshida: That's true; the trend in AI adoption is focused on "productivity improvement," essentially boosting the bottom line. However, we believe we are entering an era where generative AI will be used to create new added value, expanding the so-called top line.

And to expand the top line, a consumer perspective is essential. Since our main business is B2B, we felt we needed expertise in the consumer perspective. That's why we approached Dentsu Digital Inc., which possesses expertise in experience design and service development incorporating the consumer perspective, and Dentsu Inc.

Yamamoto: The connection originated because Dentsu Digital Inc. was already participating in Hitachi's open innovation program, the "Lumada Alliance Program."

Namikawa: As Mr. Yoshida mentioned, the strength of Dentsu Inc. and Dentsu Digital Inc. lies in designing experiences and services that capture consumer insights. Beyond the expectation that combining Hitachi's and the Dentsu Group's strengths would drive innovation, our shared vision of solving social issues through generative AI perfectly aligned, leading to our collaboration.

Yamamoto: Dentsu Digital Inc. exhibited an interactive AI showcasing Hitachi City's appeal at both the "Hitachi Social Innovation Forum 2024 JAPAN" hosted by Hitachi and the "Hitachi City Industry Festival" organized by Hitachi City. What really stuck with me was when a resident of Hitachi City said, "Hitachi can transform the city with technology—that's truly amazing."

While I often work focused on digital spaces, collaborating with Hitachi made me appreciate anew the value of creating things that touch people's lives and have a tangible, physical presence. Even within the theme of generative AI utilization, I want to explore creating such physically grounded things together with Hitachi.

Related Article:Can We Give AI a Soul? A New Challenge with Hitachi on "Regional Revitalization × AI"

Generative AI as a Trigger to Flip the "Switch of Imagination"

──What is the goal of "AI for EVERY"?

Namikawa: We aim to use generative AI to generate exciting ideas that solve societal challenges. To elaborate, we want generative AI to expand consumers' creativity or create opportunities for them to express their creativity.

──What specific services are available?

Namikawa: The first initiative we announced under "AI for EVERY" is "Today's Whimsical Recipe." This solution leverages generative AI to tackle the social issue of "food loss" from the consumer's perspective.

Specifically, it predicts ingredients likely to become surplus based on factors like the day's weather. It then generates recipes using those ingredients, tailored to the current situation, and proposes them to consumers via apps or digital signage with a "How about this recipe?" approach. This boosts ingredient sales and tackles food loss.

Yoshida: The concept for "Today's Whimsical Recipe" originated from the idea that generative AI could be used to reduce food loss, a social issue close to consumers' lives. We envision client companies like supermarkets and other retailers who sell ingredients to consumers.

──Why was "Today's Whimsical Recipe" chosen as the first initiative?

Yoshida: There are several reasons. First, Hitachi has established the "Hitachi Retail Supply Chain Vision" for 2035, outlining scenarios to contribute to the future of retail. This vision includes the goal of "Zero Waste Loss."

Above all, it was the ease of combining the solutions offered by the three companies. Hitachi has developed demand forecasting solutions and has a track record of implementation in distribution and retail sectors like supermarkets. Therefore, Hitachi handles the demand forecasting and inventory data management. The Dentsu Group handles the part where recipes are generated by referencing that inventory data and consumer data, and then displayed on digital signage in the sales floor. This led us to the idea that such a combination could work.

Yamamoto: To summarize, the reason we planned "Today's Whimsical Recipes" as the first initiative was that it aligns with our vision of

• aligns with the vision of "solving social issues with generative AI."
• and could be realized by multiplying the solutions already held by the three companies.
• and it enables approaches not only for client companies but also for consumers.

We've reached the point where we can move forward as long as client companies express interest.

──Regarding the ingredients for recipe generation, a remark made at the April 23rd "AI for EVERY" media briefing left an impression. Yamamoto-san said, "Today is April 23rd. For example, by incorporating external elements like the date, couldn't AI generate recipes based on wordplay like '423 = Shijimi Day'?" (laughs).

Yoshida: Exactly—this very "shijimi" connection was a perspective Hitachi didn't have (laughs). Seeing Yamamoto-san create something from such a small insight made me think, "Ah, that's one way to put it." It's a small thing, but even that alone made me feel the synergy of the three companies working together.

Yamamoto: I didn't prepare any material today (laughs). Before Dentsu Digital Inc., I ran a company called Data Artist. We named it that to express our intent to layer "art" onto data. That is, it's not just about science—I still hold the belief that we should be able to layer "something quirky that might never occur to us again, but suddenly pops into our heads" onto data. I think that's why the "shijimi" angle emerged.

Namikawa: Exactly—it's about the future relationship we want to realize between AI and humans. Like the shijimi example, generative AI creates "triggers" that spark consumers' imagination. Creating an "experience" that makes people want to cook something they hadn't planned on until they reached the supermarket—that's precisely the kind of exciting generative AI application we envision. It's about generative AI expanding consumers' creativity.

This time, we successfully connected the mechanisms of ingredient distribution with the creative potential of generative AI. I think it became a very clear example of the "solutions that solve social issues using generative AI" we aim for.

Yamamoto: One element generative AI brings is "personalized experiences for each individual consumer." That often seems like a closed, individual experience, right? But actually, it means "it can adapt how it communicates based on the recipient," which is also a strength. For example, telling a small child "Sustainability is important, right?" won't get through. It's better to say, "Wouldn't it be nice if the Earth stayed healthy forever and everyone was smiling?" That gets the important point across, right?

In the case of "Today's Whimsical Recipe," beyond the personalized experience generated AI provides each individual, there's a clear overarching theme: "solving food loss." Using delicious-looking recipes as a guide for shopping connects directly to solving a social issue. In other words, rephrasing it, this initiative is also about "generative AI standing between social issues and consumers, translating messages into forms easily understood by each person, and supporting solutions."

Namikawa: As a copywriter myself, I've worked on projects to "move people's hearts with a single word." But "words crafted to solve social issues" often feel a bit distant from everyday people.

So rather than just saying "Let's solve social issues," we want generative AI to translate and convey solutions—specifically "recipes" that connect to each individual's "happiness." By making many people happy through these translated solutions, we aim to contribute to society's overall happiness... That's what we're striving for this time.

When generative AI predictions miss the mark, that's where consumer-centric ideas come to life.

──What kind of feedback did you receive internally and externally after the "AI for EVERY" announcement?

Yoshida: First, the internal response was very positive. Since it enables creative proposals that Hitachi alone couldn't achieve, we've received a lot of positive feedback not only from departments related to distribution and retail, but also from those in the financial sector.

Externally, largely due to the "Today's Whimsical Recipe" initiative, we've received positive reactions across industries, particularly in distribution and retail. Many clients found the partnership between Hitachi and the Dentsu Group unexpected, but this has often led to discussions like, "Let's brainstorm together what this partnership could achieve."

Yamamoto: It might be an unexpected combination, but it would be great if it sent a broader message to other companies: "These three companies are using generative AI to solve social issues, so perhaps society will move in that direction going forward."

Namikawa: We're also receiving inquiries from both inside and outside Dentsu Inc., and we sense high expectations for initiatives unique to this three-company collaboration. Companies seeking not just efficiency gains through generative AI, but wanting to drive innovation or pursue exciting new ventures, are reaching out saying, "Let's work together." Even if you don't have a specific challenge right now, we welcome discussions starting from scratch or even casual chats.

Yoshida: That said, the more we talk with everyone at Dentsu Inc. and Dentsu Digital Inc., the more ideas we get for what we can tackle within the "AI for EVERY" framework.

For example, as mentioned earlier, Hitachi uses AI for various "predictions"—traffic congestion forecasting, crowd flow forecasting, railway passenger boarding/alighting forecasting, power transmission/distribution forecasting, and failure prediction. But AI doesn't always tell the whole truth. Even with demand forecasting, we need to start from the premise that "AI sometimes misses the mark." And that's where the Dentsu Group's generative AI comes into its own. In other words, your perspective isn't "it's unusable because predictions miss," but rather "solving the impact of prediction errors from the consumer's viewpoint."

While many remain skeptical about generative AI adoption, I believe this perspective opens up countless industries we can propose solutions to within the "AI for EVERY" framework.

──Specifically, how does the Dentsu Group's generative AI come into play when predictions miss the mark?

Yamamoto: For example, if AI predicts traffic congestion but the actual traffic deviates from that prediction, we could create solutions within "AI for EVERY" that suggest ways to stay comfortable even in traffic jams.

Recently, we had a discussion with the prompt: "Please calm a driver stressed by traffic jams during a trip in one minute." The generative AI instantly proposed a flood of ideas like "suggesting alternative routes," "speaking in the voice of the driver's favorite celebrity," or "singing a song."

If the time spent in the car becomes enjoyable, traffic jams won't seem so bad. Couldn't generative AI suggest such communication opportunities?

Yoshida: Absolutely, I think AI could be used in exactly these kinds of situations. As autonomous driving becomes more common, cars will increasingly incorporate entertainment features. It stands to reason that AI integrated into in-car systems will become more prevalent, acting as conversation partners... You know, when I talk with the folks at Dentsu Inc. and Dentsu Digital Inc. ( ), ideas like this just keep flowing out (laughs).

Generative AI becoming that "third companion" distinct from family, friends, or colleagues!

──Could you share your outlook on future generative AI applications?

Yoshida: Today's discussion focused mainly on B2B and B2C applications of generative AI, but eventually, I believe initiatives targeting employees—B2E or B2B2E—could also be part of "AI for EVERY." Some employees might feel uneasy or resistant toward AI, but generative AI could serve as a partner for them. For instance, it could provide interfaces tailored to what experienced engineers desire.

Additionally, Hitachi is already utilizing generative AI for "technology succession." As veteran technicians and engineers gradually retire, we're teaching generative AI about maintenance methods for manufacturing equipment and programming languages like COBOL, which have fewer skilled practitioners. This enables new employees to smoothly take over responsibilities.

Namikawa: Indeed, initiatives to encourage employee use of generative AI—the B2B2E domain—seem poised for expansion. The Dentsu Group is also researching ways to address discomfort or resistance toward AI. For instance, we're developing solutions that leverage character-themed generative AI to speak and foster familiarity among employees, promoting adoption from a psychological perspective.

Yamamoto: To get diverse people to actively use it, the "timing when AI approaches users" is crucial, right? For instance, during golf, offering conversation topics when players are moving and struggling to find things to say. Or having it handle proxy responses when expense reports need filing—starting with such moments. That way, people might think, "Hey, you always seem to pop up when I'm stuck," giving them a chance to reevaluate generative AI (laughs).

Honestly, I think it's vital for people to genuinely feel that generative AI makes their lives more convenient or helpful. Currently, generative AI is still limited to use by a select few, and it's not widely perceived as something that truly aids daily life or tackles societal challenges. I hope people will start using "AI for EVERY" to begin with these small, tangible improvements – like making commutes a bit easier or reducing overtime slightly.

Generative AI can provide personalized consultations tailored to each individual, and I believe it can contribute to personalized happiness. In fact, we've optimized ChatGPT for ourselves and use it to the max every day (laugh).

Namikawa: It's also evolving rapidly as a "wall-bouncing" partner, right? I recently came up with a term: you know how we have First Place, Second Place, and Third Place to describe layers of places where people feel secure? Third Place specifically refers to a space for "the other side of yourself." Drawing from that, I call the generative AI I use daily my "Third Companion."

You know how you can confide in AI about things you can't tell family, friends, or colleagues? But that doesn't mean AI replaces family or friends, or takes on the role of a colleague. That's why it's a third companion.

Yoshida: Ah, that's a great way to put it (laughs).

Namikawa: It would be great if we could view AI as a companion that exists separately from our existing human relationships—someone you can talk to when you're in trouble, maintaining that comfortable distance. I believe the technology of generative AI itself has already reached that point.

Yoshida: Hearing you talk just now, I realized that because generative AI is a trend, there's often this sense of obligation, like "we have to implement AI." Instead, if the discussion about AI being "something that supports each individual" spreads a bit more, I think we could discover a different kind of value, like the direction "AI for EVERY" aims for.

To achieve that, we want to prove that "generative AI can genuinely solve societal challenges" by deepening our collaboration. And it would be wonderful if we could involve our customers in these discussions—or even be drawn into them by our customers—as we engage in these conversations.