"Conversations" with AI Create Demand! What is the "VPA" that Moves People?

Last time, we examined AI and the advertising industry, proposing a return to the role where advertising excels: "creating demand (desire) itself." This time, we propose a model to further transform this into a service. What is VPA?

Emotion-Driven: A New Form Enabled by IoT

So, how can we use technology to realize an "external demand creation model"? IT consultant Hiroshi Irie defines three models: the "Demand-Driven Model," which addresses user needs (demand) after they become apparent; the "Event-Driven Model," which captures human behavior at the latent stage before needs are clearly defined; and the "Emotion-Driven Model," made possible by new IoT-enabled data from various sensors, such as biometric information and brainwave data (※7). Applying this framework, the "intrinsic demand-based supply optimization model" based on past demand corresponds to the "demand-driven model," while the "extrinsic demand-based demand creation model" that generates new future demand aligns with both the "event-driven model" and the "emotion-driven model" [Figure].

Furthermore, Mr. Irie states that since memories triggered by actions or emotions generally stick better with people, ① "methods appealing to events or emotions" become crucial, and ② building "mechanisms capable of sensing actions or even emotions before demands are formed" will be increasingly important going forward. This paper proposes a new agent model for the advertising industry to realize the part corresponding to mechanism ② via IoT, and the part corresponding to method ①, along with mechanisms to create emotions and desires themselves, via AI.

What is VPA? ~ The AI Interface Platform Companies Are Pursuing

Omnicom-affiliated media agency PHD held a seminar titled "Sentience: The Coming AI Revolution" at Cannes Lions in June 2015, followed shortly by the publication of a book of the same name (※8). The central AI model for marketing presented there is the VPA (Virtual Personal Assistant). VPA is a collective term for services like Siri (Apple), Google Now (Google), Cortana (Microsoft), and Alexa (Amazon). Major platform companies are all investing heavily in them as the next-generation communication interface (i.e., as media). The defining feature of VPAs is their orientation toward "voice" communication using natural language, driven by the rapid evolution of AI technologies like machine learning and deep learning. A vision of this evolution is likely the general-purpose AI OS "Samantha" featured in the movie 'her'.

Samantha is a virtual personal assistant without a physical body, making her a quintessential VPA. Though existing solely as a "voice" without even an image, she interacts with the protagonist Theodore through that voice, moves his "emotions," and Theodore falls in love. The key point is that regardless of whether an AI that behaves humanly truly possesses emotions, it can evoke human desires and induce certain emotions or actions. In the real world, discussions about communication between humans and programs possessing pseudo-personalities have a long history, dating back to the conversational program ELIZA developed by Joseph Weizenbaum at MIT in 1966. Since then, programs capable of simulating such pseudo-personalities have been called "artificial non-brains." The power of AI to "influence" people is already formidable. Even in ELIZA's era, Weizenbaum felt threatened by how quickly humans and computers could engage in emotional exchanges. Today, it's so potent that Japan's AI Society Ethics Committee discusses the potential misuse of AI that appears to possess a mind, such as exploiting it for in-app purchases in social games. In the current LPO (Landing Page Optimization) of internet marketing, where conversion is key, VPAs could potentially handle a series of activities—recommendations, Q&A, contracts—through natural language dialogue, automatically and powerfully driving sales efforts. While misuse would be problematic, its covert application by governments or military is far more alarming. I believe democratic application of this technology is possible if private entities like the advertising industry advance it transparently under agreed-upon regulations.

So, if VPAs become the next-generation marketing interface as PHD suggests, how should the advertising industry engage with them? Currently, Siri and Google Now's advantage stems from Apple and Google providing smartphone (including tablet) operating systems, enabling them to leverage smartphone sensors and data effectively. Realistically, the sensors individuals currently use (like GPS) are mostly confined to smartphones, and outputting that information (like checking your current location on a map) is also primarily done via smartphones. In other words, both the input and output of sensor information are completed on smartphones. Furthermore, many of the various IoT tools currently on the market, even if they have sensors within the tools themselves, place a strong emphasis on linking with smartphones. Currently, "IoT = smartphone," and since data aggregation is completed within the smartphone ecosystem, companies like Apple and Google hold a significant advantage (especially since data integration with cloud-based calendars and email services is also possible). However, this smartphone-dependent ecosystem is not truly "virtual" because smartphones are physical devices. Moreover, precisely because smartphones are such outstanding devices, they present an "innovation dilemma."

※7: Impress IT Leaders, Hiroshi Irie , "The Third Driver Guided by IoT is 'Emotion'" (July 6, 2015) ( http://it.impressbm.co.jp/articles/-/12539 ), September 24, 2015

※8: PHD, Sentience: The Coming AI Revolution and the Implications for Marketing (PHD, 2015)