How do we Manufacture New Knowledge Systems?

How is it possible for a man to run a marathon with just one lung?

In 1957, Donald Cantrell was diagnosed with a severe abscess, leading to the removal of his left lung. The procedure meant living with just one lung, but surprisingly, Donald has run marathons, competed in triathlons, and broke world records.

It was possible because of Compensatory Hyperinflation, a phenomenon where the remaining lung expands to fill the space left by the removed lung compensating for the loss.

Surprisingly, this concept of one part compensating for another is a universal phenomenon. Native pollinators make up for the decline in honeybee populations, children often bear the emotional weight of generational trauma, government subsidies make up for crop loss, etc.

We are witnessing a similar dynamic playing out in the AI industry. With the innovation almost reaching its saturation point that all you see around are GPT wrappers, infrastructure protocols, and AI Ethicists with questionable choices, the “Eureka” moments have become rare sights.

But who is making up for this stagnation in the AI industry?

The manufacturers of the new knowledge systems.

But before we dive into “how?”, let’s explore the 3 categories of knowledge players!

The Knowledge Players

Knowledge players are individuals or entities that contribute to the creation, dissemination, and application of knowledge within an organization/industry or world in general.

A-Priori Knowledge Players

A-priori knowledge players believe that knowledge is independent of experience. It is gained through reasoning or logical deduction. For example, 2+2=4, All bachelors are unmarried men, etc.

You would often find these rationalists and foundationalists building from the “first principles”.

A-Posteriori Knowledge Players

A-Posteriori knowledge players believe that knowledge is dependent on experience or empirical evidence. For example, scientific knowledge tells water boils at 100°C at sea level, sensory perceptions tell the sky is blue, etc.

Pragmatists often focus on the practical consequences and utility of the knowledge. Empiricists who believe that knowledge comes from sensory experience often refute the possibility of artificial consciousness.

Synthesis Players of A-Priori and A-Posteriori Knowledge

The playground for the synthesis of a priori and a posteriori knowledge holds space for mixed theories. One may find everything between an idealist and a relativist. However, the most important role in building social structures is played by the post-modernist knowledge players.

Post-Modernist Knowledge Players believe knowledge is constructed through language, power structures, and social contexts. For example, the hierarchical and interdependent nature of binary opposites like good/bad, present/absent, etc., or the creation of "docile bodies" that are easier to control through internalization of surveillance, etc.

The Game

“Whoever controls language models controls politics,” says Hannes Bajohr. While this statement initially brings to mind concerns about gender and racial biases in LLMs, the threat to democracy from AI-generated content, and the issue of centralized “value lock” due to rigid AI systems, the problem runs much deeper.

Most widely used Large Language Models are predominantly trained on global languages, particularly English. Language shapes our perception, influencing how we think and interact with the world. The dominance of any language brings its own biases and politics. The political implications of controlling LLMs extend far beyond anything we know, affecting how public discourse is shaped and controlled.

For example, “I miss you” in English focuses on the individual's feelings of absence. Whereas, “Tu me manques” in French suggests a part of oneself is missing.

Similarly, “Motherland” in English connotes nurturing and care, reflecting the feeling of nationalism in the US. Whereas, "祖国" (Sokoku) in Japanese means ancestral land emphasizing roots and heritage highlighting the resistance to outside cultural forces in Japan. And "الوطن" (Al-Watan) in Arabic means the homeland emphasizing a sense of belonging and national identity.

LLMs are horrible at understanding nuanced cultural contexts that accompany a language. Maybe they are not as blind to their blindspots as humans, but it doesn't negate the dangers of language politics they propagate.

Using LLMs for tasks like translation, therefore, can often turn into a blunder for cultures engrained. We are not too far from witnessing this as the multilingual LLMs gain popularity.

As a fun exercise, try talking to GPTo in Hindi or Punjabi if you’re a native speaker.

Rewriting The Rules

We have evolved from a culture where knowledge is no longer seen as an end in itself. Most wouldn’t bother learning a new skill unless it serves them in some of the other way. Leisure is no longer seen as a necessity but a privilege.

However, this utilitarian perspective of knowledge restricts curiosity. Ask yourself what is it that you are pursuing out of curiosity and not because it has utility, and you’ll know what I mean.

The constant pressure of utility and “making sense” is taking away our originality and hence, our ability to innovate.

But that doesn’t mean that it’s the end of human creativity. Manufacturing new knowledge systems isn’t that hard after all.

Specialization in one field can make you an expert, but it often leads to missing the holistic picture. For instance, a psychologist, a nutritionist, and a dermatologist each perceive the same human differently, yet their disciplines are closely related.

This interrelatedness is universal.

Therefore, the best approach to creating new knowledge systems or fostering innovation is to apply the frameworks and methodologies of one discipline to understand and explore another discipline.

For example,

• Understanding religion can make you a better marketer as it allows you to explore fundamental human emotions deeply

• Being a high school teacher can make you a better prompt engineer because it allows you to be a good instructor

• Playing an extreme sport can make you better at financial trading as it teaches you self-regulation in the face of intense emotions

• Cooking a meal for a family can make you a better product manager as it lets you practice multitasking and understanding customer needs

Lost in the hustle culture, the industry has forgotten that transferrable skills are learned best when enjoyed. You can teach yourself how to be creative, be a problem solver, be an innovator, or anything you want.

But most importantly, you need to teach yourself to want better.

It is only when we start wanting innovation from the industry that we can achieve it. We need to start wanting more.

Do not be satisfied with mediocre outputs. Do not be satisfied with mediocre explanations. Do not be satisfied with mediocre solutions.

You deserve better than AI girlfriends, LLMs suggesting to put glue on pizza, or AI assistants not even being able to book a flight ticket.

It’s time to take matters into your own hands!

Want to start a revolution? Share this newsletter and help others wake up too!