Category
Theme

Dentsu Inc.'s project team "AI MIRAI" explores new possibilities for AI. By welcoming Data Artist Inc. as a new driving force, we will further accelerate our AI strategy. How does the Dentsu Group envision the future of AI? Two key figures—Takuya Kodama of Dentsu Inc., who oversees AI MIRAI, and Satoru Yamamoto, President of Data Artist Inc.—discuss this vision.

From left: Satoru Yamamoto, President of Data Artist Inc.; Takuya Kodama, Dentsu Inc.

Dentsu Inc. × AI: Challenging Business and Social Issues Beyond Marketing

Kodama: As an advertising and marketing company, how do we integrate AI into our business operations? We launched AI MIRAI in 2017 to address this theme, and I feel the acquisition of Data Artist Inc. (hereafter DA) as a subsidiary has significantly boosted our momentum.

Yamamoto: We recognize the technology domain as a major pillar of Dentsu Inc.'s growth strategy. We believe that if the AI MIRAI initiative becomes the first step and Dentsu Inc.'s advertising business undergoes revolutionary change, it can impact not only the marketing domain but also society at large.

Kodama: DA has been developing AI for the marketing field for some time now. What was the initial impetus for starting collaboration with Dentsu Inc.?

Yamamoto: As AI becomes increasingly widespread, we wondered if AI marketing could be applied to other areas. For example, an "engine that analyzes emotions from people's behavior." If we could quantify when people's mental states deteriorate, wouldn't that data be useful for mental health care?

When I considered this, Dentsu Inc. stood out as the company with the most data and extensive sales channels to various businesses. I strongly believed collaborating on development would be ideal.

Kodama: Indeed, I feel Dentsu Inc.'s AI business is increasingly expected to extend beyond advertising. We often get requests from companies asking, "Please help us identify where AI can be applied across our entire business." Dentsu Inc.'s strength lies not only in its advertising track record but also in its extensive connections and data, coupled with its ability to uncover fundamental business challenges.

By the way, let's briefly review our past collaborations with DA. First, we worked on optimizing digital marketing, right?

Yamamoto: That's right. We started with media-related initiatives. Examples include the TV ratings prediction system "SHAREST" and the integrated marketing platform "STADIA" for mass media and web advertising.

Kodama: We started with AI for data inference and then expanded into other technical areas like language and image analysis. Through our collaboration with Dentsu Inc., I feel DA's service line capabilities have broadened significantly. It feels like we're mutually expanding each other's possibilities.

2018: AI Shifts from "Prediction/Inference" to "Automatic Generation"

Kodama: Actually, Dentsu Inc. recently held its financial results briefing. During that, President Yamamoto mentioned AI, stating, "AI is positioned as one of Dentsu Inc.'s core data/technology strategies."

How can we use AI to optimize the advertising space and create higher value? Beyond that, how can we use AI to solve the management challenges of our various clients? I think we'll increasingly be in a position to provide solutions outside the company.

Yamamoto: I believe society is expecting that too. After all, the Tokyo Olympics are coming in 2020. By then, at the very least, advertising-related AI technology should be fully developed.

Kodama: As Dentsu Inc., we need to develop strategies and roadmaps for that. From our perspective, the existence of DA feels like "We've gained another weapon we can use!" and it's reassuring.

Yamamoto: I think we'll keep adding more weapons (laughs). Even within the AI industry, it's said that we're at a turning point for AI right now. Until now, AI could only select the optimal choice from existing options—things like "which product is best" or "what are the ratings?" Going forward, AI itself will move toward "creating" something.

Kodama: I see. Even with the current AI MIRAI, we're moving from the analysis phase into the creative auto-generation phase, where it "creates images on its own" and "generates copy." I hope we can integrate data not just digitally, but also from real-world and offline sources, use that to create something new, and leverage AI to deliver "value only Dentsu Inc. and DA can provide."

*Reference: AI Copywriter AICO Series

It's precisely the hard-to-digitize "creative" where AI comes into its own

Kodama: I believe the spread of AI will also change how business is conducted. We can expect new developments involving various clients and players.

Yamamoto: As DA, while we continue aiming for data prediction and optimization, we simultaneously hold the grand goal of "creating creative work with AI." Copywriting, video, and images are areas that are difficult to digitize, but precisely because of that, they are AI's strong suit. We want to figure out how to enable their automatic generation.

Kodama: In the near future, we might even see "AI creating TV commercials!" If that happens, the skills demanded of creators will likely transform completely.

Yamamoto: Haven't Dentsu Inc. creators changed over the past year?

Kodama: Initially, there were concerns within the company that "AI might take creators' jobs." But now, most people think we should "explore what it means to be a creator in this new era."

Furthermore, our in-house "AI Creators Club" has launched. Several teams are operating simultaneously—like the AI copywriter development team and the team designing a Q&A chatbot.

It's not a linear shift where creators' jobs move to AI. Instead, the vision is evolving into one where AI and creators collaborate, each enhancing their respective areas of expertise.

Yamamoto: I believe we must evolve AI logic in tandem with the changes in creators. We've moved beyond digital marketing that merely ingests big data; we're now entering a phase where we train AI to learn from creators' sensibilities.

For example, the AI copywriter development team is training AI on excellent copy, but we can't ask a copywriter to produce a million pieces of copy. That's why we need to develop AI that can "extract the essence of great copy from just 100 examples."

Kodama: That's what we call "transfer learning." After learning extensively to grasp the "tricks of good copy," it applies (transfers) those tricks to different projects, selecting the best from a smaller dataset. It's not just about evolving creativity to adapt to the AI era; it's about evolving AI based on the creativity demanded.

Yamamoto: I think evolving from both angles is crucial.

Kodama: Achieving this evolution would fundamentally transform the creative profession itself. Traditionally, the workflow involved meeting with clients to understand their needs, returning to the office to create comps, and then presenting them. But perhaps we could soon "create comps on the spot, present them, and make adjustments immediately."

I feel this won't just change work styles but the business model itself, shifting from a "commission-based business" to a "collaborative creation business."

Furthermore, this could enable people outside the creative field to work in creative domains. For example, for routine projects, business producers could easily make quick creative proposals.

This would fundamentally transform workflows. When developing AI, I believe it's crucial to constantly visualize how this will reshape the entire business landscape.

World-class development advancing from our base in Mongolia!

Kodama: Moving forward, DA plans to establish a development base in Mongolia as Dentsu Inc.'s AI solutions development unit. Many might wonder, "Why Mongolia?"...

Yamamoto: That's right. First, the country's economic growth rate is incredible. In just over a decade, its GDP has increased roughly tenfold.

Kodama: Even Singapore only averages about 8% annually.

Yamamoto: Moreover, their foundational academic skills are exceptionally high. It's common practice for schools to have math clubs within their extracurricular activities, and to gather the top-scoring students from these clubs to form advanced classes.

Kodama: That's amazing.

Yamamoto: Amidst this, the country is now actively steering its education policy toward AI, saying, "Not just math, but AI too."

Furthermore, Mongolia is actually a resource-rich nation. We want to challenge this unprecedented endeavor: using AI technology support from Japan to advance Mongolia into a developed nation.

Kodama: That's a new kind of international collaboration. I hear cooperation with Mongolian engineers is also progressing rapidly.

Yamamoto: Exactly. About 20 DA engineers are coming from Mongolia. Top engineers constantly think about how to reduce their workload and streamline repetitive tasks. They don't just keep sweating away; they continuously build up things that can be generalized. I think Dentsu Inc. employees can learn a lot from this mindset and approach.

For the Mongolian engineers, Dentsu Inc.'s "thoughtfulness" and "sales mindset" will likely be stimulating. After all, the way engineers operate is fundamentally different from how producers or salespeople work. I'm really looking forward to seeing how differences in nationality, job type, and corporate culture will positively impact both sides.

Kodama: We also want to keep throwing interesting challenges their way so they don't get bored with us.

Yamamoto: Thank you. Since the distance will be significantly reduced, we aim to respond immediately to even the smallest requests. We'll focus on rapid speed and running many small PDCA cycles.

Kodama: In AI development, accepting the "unpredictability inherent to AI" is essential. It won't always learn exactly as we expect or achieve the precision we aim for. So, we must accept that "failure is also an outcome" and push forward with a sense of urgency.

It won't work unless we embrace high output and high failure rates. I believe how quickly we can move will also become a competitive advantage. Advancing projects in small, swift steps is crucial.

Yamamoto: That's right. I believe this will enable not only faster creative work within Dentsu Inc. but also quicker proposal development for clients.

Kodama: That's something to look forward to. I have a feeling we'll be able to generate new values we couldn't have discovered on our own. It would be wonderful if, amidst our different nationalities, job types, and cultures, we could provide each other with great stimulation.

Was this article helpful?

Share this article

Author

Yamamoto Satoru

Yamamoto Satoru

Dentsu Digital Inc.

Studied artificial intelligence (AI) under Professor Yutaka Matsuo at the University of Tokyo. Founded Data Artist Inc. in 2013, which merged with and joined Dentsu Digital Inc. in 2023. Utilizes AI and big data to provide numerous digital marketing services, including automated ad generation, ad effectiveness prediction, CRO, and SEO. Frequently appears on media outlets such as TV programs and speaks at seminars for companies and universities. Major publications include "How to Create Selling Logic" (Sendenkaigi) and "AI × Big Data Marketing" (Mynavi Publishing).

Kodama Takuya

Kodama Takuya

Dentsu Group Inc. / dentsu Japan

After working as a client-facing producer for digital platform companies, he has been promoting the use of AI both within and outside the company since 2018. He is currently affiliated with Dentsu Group Inc., where he is involved in the AI and technology strategy for the entire Dentsu Group, encompassing not only Japan but also overseas operations.

Also read