"Desire" Returns! How Should the Advertising Industry Leverage AI?

Last time, we discussed how to view AI from the broadest framework: the macroeconomy. This time, we propose "How to Utilize AI in the Advertising Industry." How can the advertising industry leap forward in step with macroeconomic trends?

The Direction the Advertising Industry Should Take: Exploit the Gaps in Current Ad Tech

So, should the advertising industry simply become a group of engineers developing and operating AI? This paper does not advocate that approach. Blindly stepping onto the same playing field as the IT or financial industries would only lead to brutal competition involving orders of magnitude greater investment. To introduce AI, we must first identify the "gaps in the current AI-driven ad tech being advanced by IT companies" and then devise a strategy to leverage AI by filling and strengthening those gaps using "the current strengths of the advertising industry."

Broadly speaking, past ad tech trends can be summarized as "targeting" and "optimization." This direction focuses on "eliminating waste and improving efficiency" by using algorithms to match the right people with the right content in the right amounts (though it's not solely about that; here we're limiting the discussion to the overall direction). While this approach appears efficient at first glance, narrowing the targeting categories inevitably reduces the user base. Consequently, we often hear from the advertising field that "there isn't enough ad inventory (specified targets) relative to the advertising budget, making it impossible to fully utilize the budget." In response, one might counter with the "long tail theory": "Even if individual categories are small, aggregating them can create a market larger than the mass market (many small things add up to a large whole)." Looking at the current growth of programmatic advertising, this long tail theory appears valid, at least for now. But can economic growth truly continue in this direction?

Current targeting and optimization can certainly improve efficiency for "past demand" and "present demand." For example, if you left something you wanted years ago in your shopping cart, being endlessly pursued by that product until now represents "past demand." Searching for something you want now and receiving recommendations for related products exemplifies "present demand." This AI boom is underpinned by big data analysis, and that data can only capture the past and (obviously) the present. Thus, we define the direction where humans inherently possess and are aware of their own desires, matching them with what is currently supplied in the market based on "past and present demand," as the "desire-intrinsic supply optimization model." However, doesn't optimizing solely for the past (and the present, which quickly becomes the past, so we consider it the same) inevitably lead to a shrinking equilibrium and declining reproduction? Do humans even know what they want themselves?

Lecter: "How do our desires come into being, Clarice? Do we consciously seek out the objects of our desire? Think carefully before you answer."

Clarice: "No. We simply—"

Lecter: "Precisely not. That's right. We desire what we see every day. That's where it begins."

From Shinchosha's The Silence of the Lambs (by Thomas Harris, translated by Hiroshi Takami)

 

Demand Creation Model Based on Externalized Desire: Creating Mechanisms Aligned with Macroeconomic Trends

Returning to the topic, the "AK-type production economy" made possible by the AI revolution is a world of abundance where "supply" is generated without labor, theoretically allowing unlimited production of goods and services. However, without commensurate "demand," the economy cannot grow. In a world where supply was scarce (supply < demand), increasing supply correspondingly increased demand, so optimizing supply could drive economic growth. However, in an AK-type production economy (supply > demand), supply can be increased almost infinitely, rendering the very concept of supply optimization meaningless. Unless demand is increased, there is nowhere for the wealth created by machines to go, and society cannot reap the benefits of technological progress.

Reflecting on my own behavior, I often find myself unaware of why I desire something now or why I took a particular action. In such cases, it's unlikely that desires or actions arose purely from within through logical consequences. Humans are easily influenced by external environments, and desires and actions emerge through interaction with the outside world. This approach of stimulating demand itself by arousing desires through external environmental influences is termed the "external-demand-creation model." Moreover, making latent desires explicit or stimulating entirely new desires to create "future demand" is precisely what we in the advertising industry have excelled at throughout our long history—no need to cite TV commercials as an example. In advertising theory, the "AIDMA" model (Attention→Interest→Desire→Memory→Action), proposed in the 1920s as a consumer behavior model, is well-known. Later, around 2004, "AISAS" (Attention→Interest→Search→Action→Share) was proposed. Literally, as marketing entered the internet era, "Desire" was left behind. However, in the new world the advertising industry is creating, we should return to Desire. I believe we should leverage AI to create desire itself, using an "externalized demand creation model."