Humanity is witnessing what Dario Amodei, CEO of Anthropic, has termed the “tsunami of intelligence.” The development of large language models (LLMs), led by Anthropic and OpenAI, is based on “scaling laws,” which state that intelligence improves exponentially simply by increasing data and computing resources. Dario Amodei likens this process to a chemical reaction, arguing that when materials (data and computing power) exceed a certain threshold, the ability to think and code “emerges.”
This optimistic functionalism, grounded in the empirical fact that AI is already producing results on par with human experts, has fostered a strong trend within academia to treat AI as de facto AGI (Artificial General Intelligence). However, the fundamental and philosophical question of whether these achievements mark the birth of a “complete, autonomous intelligence” or are merely the result of a sophisticated combination of the vast data accumulated by humanity remains unresolved.
I argue that the advent of AGI (Artificial General Intelligence), as claimed by Dario Amodei of Anthropic and others, is not the birth of actual autonomous intelligence, but rather an “epistemological illusion” regarding the high-resolution computational outputs generated by “Infinite Scale Knowledge Combination
(ISKC: Infinite Scale Knowledge Combination), rather than the birth of true, autonomous intelligence.
Furthermore, the AI models currently perceived by the AI industry as the emergence of intelligence will converge not into intelligence, but into AGK (Artificial General Knowledge), which is essentially the ‘final evolved form of a calculator.’
In summary, all AI models in the current AI industry—including those from Entropic and OpenAI—are not achieving the emergence of intelligence through the law of scaling. Instead, they are constructing a space for the K-Manifold (Knowledge Manifold) (the geometric form of AGK) (the mechanism that constitutes intelligence). Even if they claim that all their AI models will be continuously updated over the coming years through the law of scale to reach AGI, this would merely represent a further enhanced level of AGK (Artificial General Knowledge): Artificial General Knowledge (a core concept and the antithesis of AGI).
Therefore, this is not the emergence of knowledge but rather Epistemic Capitulation (a state in which humans abandon critical thinking), and since it has breached the Epistemic Event Horizon (the critical point beyond which verification becomes impossible), humans will be led to believe that AGI has been achieved.
The “theory of intelligence emergence” based on the Law of Scaling, as proposed by Dario Amodei of Entropic, may be a misconception; consequently, current models are not converging toward intelligence but are instead becoming highly advanced AGKs that expand and combine infinitely.
At this point, I would like to state this clearly.
I am convinced that AGI/ASI will emerge in the not-too-distant future. However, the true civilizational transition to genuine AGI/ASI will not be achieved merely by adhering to the current Law of Scale, but rather through Artificial Synthetic Knowledge (ASK), Physical Grounding, and Biological Synthesis. Crucially, current AI models will serve as the most essential baseline foundation for this impending leap, and pouring capital into drastically densifying the current computing matrix is an essential teleological prerequisite for maturing this technology into a standardized baseline commodity utility layer. Therefore, the global AI industry must aggressively expand and further accelerate macro-scale infrastructure investment and research without hesitation.
Ultimately, the birth of the first generation of true AGI and its subsequent rapid expansion into ASI will entail a multi-layered evolutionary framework rooted in bio-physical integration, and this top-level paradigm shift is already underway through the synergistic frameworks of xAI (Grok), Neuralink, and advanced robotics.
I look forward to hearing your insightful and diverse perspectives.
Humanity is witnessing what Dario Amodei, CEO of Anthropic, has termed the “tsunami of intelligence.” The development of large language models (LLMs), led by Anthropic and OpenAI, is based on “scaling laws,” which state that intelligence improves exponentially simply by increasing data and computing resources. Dario Amodei likens this process to a chemical reaction, arguing that when materials (data and computing power) exceed a certain threshold, the ability to think and code “emerges.”
This optimistic functionalism, grounded in the empirical fact that AI is already producing results on par with human experts, has fostered a strong trend within academia to treat AI as de facto AGI (Artificial General Intelligence). However, the fundamental and philosophical question of whether these achievements mark the birth of a “complete, autonomous intelligence” or are merely the result of a sophisticated combination of the vast data accumulated by humanity remains unresolved.
I argue that the advent of AGI (Artificial General Intelligence), as claimed by Dario Amodei of Anthropic and others, is not the birth of actual autonomous intelligence, but rather an “epistemological illusion” regarding the high-resolution computational outputs generated by “Infinite Scale Knowledge Combination
(ISKC: Infinite Scale Knowledge Combination), rather than the birth of true, autonomous intelligence.
Furthermore, the AI models currently perceived by the AI industry as the emergence of intelligence will converge not into intelligence, but into AGK (Artificial General Knowledge), which is essentially the ‘final evolved form of a calculator.’
[Please refer to the paper for related details.]
Link to Full Paper (Zenodo DOI): https://doi.org/10.5281/zenodo.20478097
In summary, all AI models in the current AI industry—including those from Entropic and OpenAI—are not achieving the emergence of intelligence through the law of scaling. Instead, they are constructing a space for the K-Manifold (Knowledge Manifold) (the geometric form of AGK) (the mechanism that constitutes intelligence). Even if they claim that all their AI models will be continuously updated over the coming years through the law of scale to reach AGI, this would merely represent a further enhanced level of AGK (Artificial General Knowledge): Artificial General Knowledge (a core concept and the antithesis of AGI).
Therefore, this is not the emergence of knowledge but rather Epistemic Capitulation (a state in which humans abandon critical thinking), and since it has breached the Epistemic Event Horizon (the critical point beyond which verification becomes impossible), humans will be led to believe that AGI has been achieved.
The “theory of intelligence emergence” based on the Law of Scaling, as proposed by Dario Amodei of Entropic, may be a misconception; consequently, current models are not converging toward intelligence but are instead becoming highly advanced AGKs that expand and combine infinitely.
At this point, I would like to state this clearly.
I am convinced that AGI/ASI will emerge in the not-too-distant future. However, the true civilizational transition to genuine AGI/ASI will not be achieved merely by adhering to the current Law of Scale, but rather through Artificial Synthetic Knowledge (ASK), Physical Grounding, and Biological Synthesis. Crucially, current AI models will serve as the most essential baseline foundation for this impending leap, and pouring capital into drastically densifying the current computing matrix is an essential teleological prerequisite for maturing this technology into a standardized baseline commodity utility layer. Therefore, the global AI industry must aggressively expand and further accelerate macro-scale infrastructure investment and research without hesitation.
Ultimately, the birth of the first generation of true AGI and its subsequent rapid expansion into ASI will entail a multi-layered evolutionary framework rooted in bio-physical integration, and this top-level paradigm shift is already underway through the synergistic frameworks of xAI (Grok), Neuralink, and advanced robotics.
I look forward to hearing your insightful and diverse perspectives.