The Real Power of AI Is Its Command of Human Language

The power of language-based artificial intelligence is understood by anyone who has had writing assistance from ChatGPT or Google’s Bard. By constantly analyzing the words used across an unimaginably huge internet data pool, AI has proven remarkably adept at understanding the underlying relationships between language-based concepts and, as a result, the collective wisdom of humanity.

These concepts, it’s fair to say, represent the building blocks of human cognition, consisting of thoughts, ideas, people, places, symbols, products, intentions, hashtags—all the things conveyed within our language. In the simplest of analogies, they might be thought of as the descriptors of our memories: When we think about “diamonds,” we might immediately associate “sparkle,” “ring,” “rare” and “valuable.” We instantly bring together the memories and associations of these individual concepts within the context of “diamond.” 

How does this apply to marketing? Let’s take a workwear retail brand as an example. Think about all the different ways people of all walks of life might define this brand, the associations we’d access if we got inside their minds: We’d expand our understanding of this audience exponentially and uncover the various and elusive reasons why people care about the brand. In fact, this is how marketers have always wanted to define their brands, but they have been constrained by the limits of consumer data. 

Now imagine being able to consume and associate all publicly available information that exists in the world about this brand and its products. The concepts that define what entices consumers would emerge—everything from its association with celebrities who wore the gear to partnerships the brand has formed, to granular associations based on age, aesthetics or sustainable practices. Imagine making media buy decisions with that knowledge. How many other marketing opportunities might emerge?

This is the magic of language-based AI made plain, and the technology already exists. But it doesn’t stop with simple associations. Each identified concept, and its associated concepts, can in fact be quantified in terms of their relative importance to the marketing objective. Sentiment, frequency, recency, trend and even demographics can be assigned to these concepts, and doing so transforms them into units of knowledge that begin to approach that of memory. When actionable within media, this translates into new audience opportunities, each of which can be judged qualitatively and quantitatively. 

And it’s not just the power of AI that makes it revolutionary, it’s the accessibility of the information that trains AI—it can now read, understand, enhance and act upon these learnings in near real time without huge research budgets and large teams of analysts—and, perhaps most importantly, without the need to limit audiences to a handful of expensive and ultimately limited set of the same targetable consumers everyone else is using.

Identity-based marketing: Goodbye and good riddance

The crumbling away of the cookie, and the inherent problems of digital identifiers, are well understood by even the most junior marketer. To so many in our space, this represents a catastrophe. But the truth is, thanks to AI, it is the most liberating development in marketing in 25 years. 

Let’s set aside the ethical considerations of cookies, the mistrust consumers feel about them and the brand risk associated with using them. On a purely practical level, tying marketing to consumers is inherently limiting. Marketing should not be about identifying which individuals might buy a product; instead, the goal has always been to better understand why a product is itself interesting. 

Diamonds are interesting because they are rare and valuable. With this knowledge, we no longer limit ourselves to finding and targeting consumer profiles that are expensive to buy and growing increasingly hard to scale. Instead, we learn how to position our product or service to anyone who might find value in what is being offered. When we define our target based on who, we get more of the same whos; when we define based on why, we get all the whos that share the same why. 

This removes, at a stroke, many frustrations and limitations, including the need to compete for a limited group of profiles that drive up the cost of their eyeballs. It moves us instantly past the limitations of having to rely on any-party data. It opens up entirely new audiences to be quickly identified, located and communicated with. And it respects consumer privacy.

Though AI moves us past the age of simple demographic-based marketing, it nonetheless can be trained to understand this very data. Each concept identified by AI can be understood in terms of the people likely to find value in that concept—and we find that, just as brands are defined by concepts, so are people, with additional important and incisive insights to discover.

Ultimately, then, we have come full circle with marketing. For 25 years, the promise of digital marketing has been tying impressions back to interested individuals. Now we can ascend above that plateau. And it’s hard not to appreciate the irony of artificial intelligence being what returns marketing to its most human essence.