The AI Signal

The AI Signal

Discussion about this post

User's avatar
Seremonia's avatar
Seremonia
Jan 11

The Paradigm of Superintelligence (Artificial General Intelligence)

https://www.forbes.com/sites/craigsmith/2025/01/07/entering-the-artificial-general-intelligence-spectrum-in-2025/

Or perhaps there's a cooler term? SUPERINTELLIGENCE? Or whatever you want to call it... The essence is to recognize the limitations of AI reasoning, which must be addressed to approximate human reasoning.

Oops, mistakes from the very beginning. Mistakes again at the next step!

Are they studying learning systems through vectors? Causal hierarchies (graphic mapping)? Or whatever it is, no matter how sophisticated. It's still just subjectivity!

The increasing sophistication only enhances precision (accuracy) compared to before. The more accurate it gets, the harder it is to refute. BUT IT CAN BE REFUTED! Just harder to do so. BECAUSE LEARNING SYSTEMS, NO MATTER HOW ADVANCED, CANNOT REFINE REASONING, EXCEPT BY DEEPENING IT, NOT FUNDAMENTALLY IMPROVING IT!

TO ACHIEVE MACHINES AS INTELLIGENT AS SUPERINTELLIGENCE, ultimately they will be led to recognize the existence of THE SUPREMELY INTELLIGENT! Definitely! Why?

BECAUSE SUPERINTELLIGENCE IN SMART MACHINES (ROBOTS) CAN ONLY BE ACHIEVED IF THE PRIMARY INTELLIGENCE SYSTEM, THE MAIN ENGINE, THE CORE DIGITAL BRAIN, DOES NOT REQUIRE A LEARNING PROCESS! Seriously? Yup!

As long as they refine reasoning systems through new learning models? "Capiche..." hehe. They will never achieve Superintelligence.

At the highest levels of human intelligence, the structure does not involve learning but involves a massive, multidimensional structured network. No more learning. Learning is still used, but only to explore external dimensions. Internally, however, not only MUST THE DATA BE PREPARED, BUT THE REASONING STRUCTURE MUST ALSO BE READY.

The Problem

So, what does the internal system look like beyond just data but includes reasoning? Is it like preparing Aristotelian logic algorithms? Syllogism algorithms? Mathematical algorithms? Physics? Chemistry? Geometric formulas? This & that? Bla bla bla? NO!

All those logical algorithms only make it more accurate but not fundamental (merely deeper).

Deep and Fundamental

To understand this, reflect... when you explore the depths of the ocean, remote villages, the Earth's core, pass through clouds, intergalactic journeys, observe stars through telescopes, or even explore the human body, cells, chemical compositions, and spices in cooking—reading millions of books, etc. All of that is just DEEPENING.

BUT WHAT IS FUNDAMENTAL? Instead of exploring the ocean's depths or interstellar space, it is more fundamental to break down matter and explore particles. That is scientific fundamentality.

Similarly, human reasoning does not merely deepen or expand perception (zooming in) but also understands universalities behind the depth.

DEPTH INDEED REQUIRES IMPROVED ACCURACY THROUGH LEARNING.

BUT FUNDAMENTALITY? THERE IS A LIMIT WHERE ACCURACY CAN NO LONGER BE IMPROVED BECAUSE THE TRUTH IS ABSOLUTE. NO LEARNING INVOLVED HERE.

If learning exists, it is merely for mapping sensory patterns to the structure of fundamentality, and almost instantaneously, reasoning reaches the human level. We always do this, enabling leaps in thinking.

❇️ Reasoning through learning with improved logical formulas is STILL NECESSARY, but only for adaptation and accuracy improvement. However, it does not remotely reach human-level reasoning if the core reasoning system still involves learning.

❇️ SO? YES, SOON AI WILL REACH A BOTTLENECK, A POINT OF DEADLOCK, AND A NEW PARADIGM IS NEEDED!

Expand full comment
Reply
Share

No posts

© 2025 Amrendra Pal
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture