Does Gen AI Really Add to the Knowledge Economy?
What is it all leading to?
Background
With the recent release of ChatGPT 4o Image Generation, many have play-acted as Studio Ghibli master animators and shared with the world their Studio Ghibli-style images of their portraits of family and friends. While this play-acting trend has gone somewhat viral, NONE of those Studio-Ghibli-style images have in themselves gone viral. This leads me to wonder if this new AI image generator adds or scaffolds towards any real utility to the broader knowledge economy. Or is it just low-grade entertainment for the attention economy?
And so I dedicate my 85th article to a discourse on the knowledge utility of Gen AI.
(I write a weekly series of articles where I call out bad thinking and bad practices in data analytics / data science which you can find here.)
Knowledge Economy
Let’s start with the basics. What exactly is meant by the term, the knowledge economy? All economies are defined by the matching of demand and supply. And in this case, the knowledge economy is defined as the consumption (i.e. demand) and production (i.e. supply) of intellectual capital. Intellectual capital is further deconstructed as cognitive-based (i.e. mental) intangible assets such as proprietary information, knowledge, and skills that can be monetised for economic benefits.
In data analytics, we define knowledge as “information that leads to decision-making”, i.e. information that has consequence. Information itself is defined as data that has interpretative meaning, i.e. data that is contextualised. I wrote about this in one of my earliest article “Information is NOT in the Data” and in a later article “What’s in a Name — the Problem with Roles & Sub-Domains in Data Analytics”.
Utility Continuum
One can look at the contribution to the knowledge economy through what I call the Utility Continuum:
- knowledge creation → knowledge dissemination → knowledge application.
It should be obvious that utility increases, perhaps even exponentially, from creation to application.
It is generally well-established that predictive AI plays in the “knowledge application” phase by supporting better and faster decision-making through identification and automation of useful information. Generative AI, on the other hand, plays largely in the “knowledge dissemination” phase. It reduces the friction in the “seek & search” of known information and knowledge. Both types of AI do not create knowledge; this is still very much a human-based endeavour today and the foreseeable future, until the advent of some kind of AGI (artificial general intelligence) embodied in a robot.
With the new ChatGPT image generator’s ability to mimic the artistic style of famous artists, what grand utility does it create? Does that capability operate in the knowledge dissemination or knowledge application phases? As it stands, I think neither. The mimicry of well-recognised artistic styles isn’t disruptive. It isn’t even really displacing. It is, in fact, simply devaluing. There is experience utility associated with art. Good art causes a shift in perspective, and in so doing, can sometimes lead to the creation of new knowledge. And that is its role in the utility continuum. But there is arguably no experience utility associated with mimicry of art. It doesn’t disrupt the industry, and it won’t reduce the number of artists in the marketplace.
Knowledge Dissemination vs Knowledge Democratisation
AI image generation, outside of mimicry, will and has displaced commercial graphics & design agencies. And that is to be expected. The simple, low utility graphics & design stuff will get automated away through empowerment of the masses via templates, like with every technology that came before. But great graphics & design will still remain in the hands of domain experts (trained in the artistic discipline), who will now utilise the AI image generation tool to increase their job bandwidth. Unit cost will come down, but utility will persist.
And this is perhaps the greatest revelation of the Gen AI movement — that knowledge dissemination doesn’t necessarily lead to knowledge democratisation (i.e. everyone being able to gain the same knowledge expertise). Rather, Gen AI’s utility and contribution to the knowledge economy will be realised through the amplification of those who already have the necessary domain knowledge.
Conclusion
The Studio Ghibli mimicry trend will die out. Really fast. Beyond crass entertainment value, it has no real utility. But the trend speaks to the greater distortion that Gen AI is going to be a game-changer disruptor in the near-term. While the technology is impressive, if its place on the utility continuum and contribution to the knowledge economy is ill-defined, it’s just not really going to matter.
