The only purpose of general AI. The only one that I can fathom

#lang/eng

What are the biggest inventions of humanity?

Not so long ago I was writing about the most influential scientific work, the Green book and the impact it had on Schrödinger and his ability to produce the book “What is life”1, which later turned out to be the source of inspiration for Watson and Crick in their search for DNA’s double helix. This got me thinking2 what we can single out as the most consequential invention? The wheel? The steam engine? The Internet? The AI?

No, no and no. To find the answer we have to go back, back and back in time. What invention granted humanity every boon it got, from the wheel to modern AI? What made possible these inventions and many others? Surely, we can say that our brain allowed it, or we can be more cunning and say the ability to interact socially made possible every breakthrough we had from steam engines to rockets. But I would not be so sure about it. If we are hard-headed enough in our pursuit of finding the origin of “it all”, we might find ourselves once again in Ancient Greece in front of Aristotle3 who first4 attempted to give existing sciences their structure5, classify them and by doing so created a precedent from which the scientific method6 emerged.

Anecdote

The irony of me reaching Aristotle lies in the same distrust of the inductive method applied to the scientific method that I share with him. Aristotle, like me, favored deduction over induction; he did it via the syllogism, which nowadays I don’t much like due to its rigidity. I am wary of treating induction as the thinking that yields certain knowledge.

The path that Aristotle paved was later trudged by Ibn al-Haytham (~1000 CE), Francis Bacon (1620) and Galileo, who possibly created the greatest boon of humanity: the discovery (invention) of the scientific method7, which firmly stands on our ability to think in a specific manner.

Pace of discoveries

Before going deeper into “the only purpose of AI” I wish to highlight another phenomenon: the pace of discoveries humanity has been able to produce since Aristotle’s classification effort.

It has slowed and in some disciplines completely halted8.

2026-06-08T10:03:41.708709 image/svg+xml Matplotlib v3.10.0, https://matplotlib.org/ More effort — fewer breakthroughs Across the 20th century the inputs to science kept climbing while the novelty of the output kept falling. INPUTS ▲ what we put in ×18 more researchers needed to sustain Moore's Law than in the early 1970s Bloom et al. 2020 +6 yrs rise in the average age at 'great invention' across the 20th century Jones 2009 1930 1950 1970 1990 2010 0 25 50 75 100 Research productivity of the U.S. economy new ideas per researcher (index, 1930s = 100) Down 41× since the 1930s (~5%/yr) 1945 1960 1975 1990 2005 0.0 0.1 0.2 0.3 0.4 0.5 Disruptiveness of new research papers average CD5 index (1 = field-shifting, 0 = purely consolidating) −100% −92% The 1953 Watson–Crick DNA paper sits near the top of this scale — the kind of field-shifting paper that has become rare. Social sciences Physical sciences Sources: Bloom, Jones, Van Reenen & Webb (2020), “Are Ideas Getting Harder to Find?”, Am. Econ. Rev. 110(4). Park, Leahey & Funk (2023), “Papers and patents are becoming less disruptive over time”, Nature 613. Jones (2009), “The Burden of Knowledge…”, Rev. Econ. Studies 76(1). Panel A is a faithful exponential at the paper's stated rate (factor-41 decline, ~5%/yr). Panel B endpoints are the reported 1945 & 2010 CD5 values; the connecting curve is schematic (decline steepest before 1970, per Park et al. Fig. 2). OUTPUTS ▼ what comes out

The scissors of modern science — the inputs to research keep rising while the novelty of the output keeps falling.9 10 11

From what I see, read and study, I think that the pace slowed down, and the reason, as best I can tell, is that the human race reached its limit in shouldering the burden of knowledge. We are on the plateau, and from here there are only two paths: downward (degradation) or an upward spiral, for which we need enhancements. We got ourselves so far by the sheer will of thought. Maybe we can make another move that puts us on an upward trajectory?

I understand that the claim “shoulder the burden of knowledge” needs some proof, so here we go: the tentative consensus is that it is not only the brain but a whole host of other factors at play, according to some publications9 10 11. Take another look at the graphs above.

From the exposition above I assume that AI might be the next big thing that current science12 needs. The lever that lifts the burden. The reason why I think so lies in another belief that I explored at length in my recent book13, which is: what does it mean to truly learn/understand something?

I think that no matter the amount of research poured into it, we still cannot understand how our brain works. We study and study, but are still far from understanding the intricate principles by which this magnificent tool that nature endowed us with works. The impossibility of understanding originates from a known phenomenon: to understand something means to recreate it14.

whatyoucannotcreateyoudonotunderstand.jpeg

whatyoucannotcreateyoudonotunderstand.jpeg

Hence, observing intelligence is a necessary but still insufficient step. We need to build “the brain” — which, incidentally, is what the science of AI15 is all about: rebuilding the brain in the simplicity of zeros and ones. Not studying the brain with MRI and EEG, but actively building a synthetic brain that could be applied to anything. Any kind of data, any problem, just as the biological one that humans use to solve almost every problem. The kind of brain that is infinitely adaptable.

Humans invented the modern world with the sheer power of thought, but maybe in the 21st century the capacity of the biological brain is not enough for scientific breakthroughs of the magnitude16 that Watson and Crick17 made?

To develop this concept further I wish to look into the structure of scientific work, or better, the structure of any intellectual endeavour.

BIO

🧠 theBrain mapping

ID: 202606080758 Source:: Friend:: Child:: Next:: 85% of any intellectual work is …

Footnotes

  1. Schrodinger E. What is Life? (Canto Classics): With Mind and Matter and Autobiographical Sketches / E. Schrodinger, Cambridge ; New York: Cambridge University Press, 1992. 184 c. ↩

  2. And to answer this we need to look into evolution of humanity ↩

  3. Who is often called “Father of sciences”. He was the first to insist on empirical observation of natural world. ↩

  4. I mean systematically ↩

  5. He created the earliest systematic classification, dividing sciences into theoretical (physics, maths, philosophy), practical (ethics, politics) and productive (poetics, rhetoric). https://plato.stanford.edu/entries/aristotle/ ↩

  6. Aristotle began and later Bacon (1620), Galileo and the rest of great minds carried on. ↩

  7. Or structure of scientific thought ↩

  8. You will have some data to look into down the line. The reports and references to publications. ↩

  9. Bloom N. Are Ideas Getting Harder to Find? № 4 (110). ↩ ↩2

  10. Park M., Leahey E., Funk R. J. Papers and patents are becoming less disruptive over time // Nature. 2023. № 7942 (613). C. 138–144. ↩ ↩2

  11. Jones B. F. The Burden of Knowledge and the ‘Death of the Renaissance Man’: Is Innovation Getting Harder? ↩ ↩2

  12. I am not sure which it might be, but I would bet my money on Biology, Chemistry and Physics ↩

  13. Агамалиев Р. Делай — и учись: как превратить обучение в управляемую систему достижения целей / Р. Агамалиев, Москва: МИФ, 2026. 192 c. ↩

  14. I don’t remember where I got this, but in my note “What I cannot create, I do not understand” is attributed to the blackboard motto of Feynman ↩

  15. Lets make it a new domain of science, shall we? ↩

  16. Level that changes our perception of reality ↩

  17. Papers that I cited above argue that DNA’s double helix had been the last significant discovery of our age. Since 1953 “nothing new”. ↩

Узел связи // Comm Station

TG: ОФЛАЙНBBS: ОФЛАЙН
Secure P2P Neural Link // Digital Garden v2.5