So, there is the “biological” and “evolutionary” convergence.
What does it mean to “know” something, and what is worth knowing?
I have seen very nice guys who know very long, elaborage mantras and can recite then verbatim. Notice that unlike sutras, which are versed human languages, matras are just rythmical sounds, which does not translate into any vocabulary or any semantic meaning.
One could, instead, by systematic repetitive exposure, memorize some concepts of very abstract branch of so-called moral or political “philosophy”, or some primitive abstract doctrine, such as Marxism or Non-Duality (Advaita). Would this be qualitatively “better” or just equivalent?
Ancient thinkers formulated the “conditioning” principle long ago:
“You are what your deep, driving desire is. As your desire is, so is your intention. As your intention is, so is your will. As your will is, so is your deed. As your deed is, so is your destiny.”
– Brihadaranyaka Upanishad
and also
“Watch your thoughts, they become your words; watch your words, they become your actions; watch your actions, they become your habits; watch your habits, they become your character; watch your character, it becomes your destiny.”
― Lao Tzu
This, of course, is socially constructed (subtly “refined” through uncountable rewrites and reformulations) paraphrase of the Upanishadic “law”, but they have intuitively captured the underlaying biological process, which we now understand at least in principle.
What is the process? Repeated exposure to a particular set of stimuluous through the sense organs or reptitive voluntary actions.
Here is exactly how every biological organism which has a brain builds and continuously maintains an inner “map” of the outside reality (which is, as it was famously stated, “not the territory”).
The underlaying fundamental principle (and the mechanism) here is that these experineces are by no means arbitrary, and reflect all the physical, physiological, and even social constraints of the shared environment, and thus, the inner “representation” (whatever it is – a “current state” of this or that particular set of synaptic gaps) eventually converges on “seeing things as they [realy] are”.
The actions with have been confirmend to align with the “expected outcomes” (even partially, and this is crucial to “keep going”) gets “rewarded” – the released hormones also act as “markers” for further “myelination” and thus reinforcement for the similar repeated behaviror in the future.
All living forms move through their shared environments in exactly the same way. Not just that, humans have learned to move about (crawl, then walk) in this very way, subject to the skeletal muscle “machinery” they were evolved to be equipped with – but they learn how to move their legs. Goats or horses, on the other hand, just walk away after being born (again, evolution at work).
These facts, by the way, show that such a complex behavior as coordinated movement (as in goats) can be both learned “from scratch” and “encoded”.
There is the catch – every single human activity, from very fine ballet movements (which require decades of deliberate practice – repeated training every single day), to sculpting, painting, any fine art in general, music composition, boxing, any physical activity whatsoever, has the very same underlying “algorithm” – progressive “rewarding” of “positive” expected outcomes, especially partial ones, so the brain’s inner representation “literally” converges to match the constraints of What Is.
Ballet training or Boxing (Muai Thai or whatever) is the cannonical and well-studied and well-understood example – they just repeatedly alter both the muscles and the brain regiouns with control and coordinate the movement, so eventually they develop the very refined structure in the brain, which “does” the actions “just right”.
Make no mistake – spending years in a gym or a training hall is as meaningless as to spend years learning some fancy abstract bullshit – it is the very same process and the very same futility.
Now we understand the principles. Let’s talk some basic sociology and economics.
Humans (and even animals!) consider some actions to be “premium”, according to their physical or cognitive “difficulty” – something has to be difficult, taking long time and a difficult to learn skill (in the case of crafts) or be exceptionally elegant, beyond ordinary ability (as in arts). Otherwise what is not rare is abundant and thus cheap.
So called “intellectual” achievements are being rated in just the same way – from mere memorization of facts, to ability to perform mathematical or logical derivations, which are, again, require a lot of repeated exposure to be come “familiar” (but in fact to develop an appropriate set of myelinated axons and synaptic gaps over time).
The current computing power of a data-center-size “frontier” model delivers an appearance of intelligence from a sheer brute-force.
Whatever “made-out-of-meat” humans spent lifetimes of forming an appropriate inner representation inside their heads do, a modern LLM emulates or appears to be capable of doing in just a few seconds.
Now what?
Yes, it operates at the level of repeated exposure to the tokens, not the meanings, but the resulting slop is good enough and outperforms 90% of humans in any written-text-based cognitive tasks.
No, it does not “understand” mathematics, but it synthetize the “answers” by selecting the next most probable token out of the most frequently used ones, and once in a while it is a “new result”, just because the inference algorithm selected less frequently used – “less common sense” token.
Consider some good guys with some carefully written niche good books (Richard Bird, Joe Armstrong, Lawrence Paulson, Xavier Leroy, etc.) These guys spent decades to form a proper synaptic structures to match particular abstract problems, and today a slop generator can produce the whole books in a few hours.
Not just that, if you know what to ask for, you can trick it, if you are exceptionally clever with words and exceptionally lucky, to produce an optimal result for any non-bullshit (grounded in the underlying reality) problem (which always, in principle, eventually converge on What Is).
Now what?
So-called “Knowledge Work”, not as in Cal Newport, who is just a narcissistic clown, but as in Simon Peyton Jones or Xavier Leroy is kinda fucked up (this is a technical term).
Yes, there is a whole emergent “occupation” of validating the slop and trying to reach an optimum by steering the slop generator in the right direction, but as you may notice, the activity is cheap, and the reward is just dopamine-serotonin oscillations.
Have I told you the metaphor I came up with when studied the Gradient Descent algorithm back in 2010? A properly folded sheet of paper can cover any mountain, achieving a perfect match. This is how the Universal Algorithm can be properly conceptually visualized.
Have you seen how after a fresh snowfall the commonly used trails re-emerge in the morning under your window in some 9 storey concrete commiblock? This is how “the brain works”. Any brain.
Have a nice day and think about “now what?”
It is not the case that mere gedonism won (but it gained a few points), but the ancient Indian esoteric traditions, which deemphasized social “success” and focused on “returning” to so-called “animal-mind state” (a pure representation, not being confused by abstract concepts made out of words) are.
Not being carried away by words, since the words now are very cheap again.
One can codition oneself to get the serotonin spikes from any repeated activity whatsoever, including writing nice books no one would read. It does not matter, after all.
The dopamine-serotonin pathways will keep you moving on, and that is Life Itself and its only “meaning”.