Embracing the Savage — That word and a different Claude.
The Afterword To Embracing the Savage brings that particular journey to a close, delves into what the word 'savage' actually means, and ends up in an unexpected place: AI, by way of a Claude who predates the Anthropic one. Read the full series here.
In this once in a lifetime journey, I spent the majority of my time in or close to wilderness. Walking, sitting, gathering, eating, listening.
It is there that masks, vital in other circumstances, began to melt away, revealing something I came to think of as “savage”. That word, I'm sure, evokes different images based on where you're from and a myriad of other factors.
Indeed, that is at the heart of the book The Savage Mind (1962) by the French anthropologist Claude Lévi-Strauss. He rejects the evolutionary hierarchy that casts so-called "primitive" thought as an inferior or prelogical version of modern thought. "Savage thought," as he coined it, exhibits systematic processes of observation, differentiation, classification and relation — not a lesser intelligence, but a different organization of intellectual materials.
"Savage thought" resembles the bricoleur, who builds a diverse number of things out of an understanding of the materials at hand, while the engineer conceives himself as dealing with the totality of available means and concepts.
The word "savage" in "Embracing the Savage" is neither the anthropological category Lévi-Strauss describes, nor a synonym for primitive. Its significance is something closer to the part of the human being that precedes, escapes, or refuses complete domestication by the structures we build around ourselves.
What, then, happens when the savage is not merely encountered in an anthropological account, but rather in oneself?
There was no grand design on my part when I embarked on the road trip described in "Embracing the Savage"; simply a sense that there was more to me and more to the world. As I acquainted myself with setting up tents on the fly, walking winding canyons, or traversing path-less sand dunes, I began to peel back layers of conditioning, training, and experience in academia, manufacturing, and business. As described in the series, I happened to be in Moab when I stumbled across the aforementioned book on sale for a couple of dollars at a used bookstore.
Perusing the tome, it gave me an insight into what I had undertaken, but it was not all-encompassing. It wasn't, because Lévi-Strauss's thesis is on the mind. As committed and open as he may have been, he was a French intellectual attempting to decode the savage experience — one which is not only mind, but also body, heart, and soul. It is here that I, a south-Indian who spent more than a decade in the West, offer a different perspective. Further, some of my formation also took place living, working, and travelling in Latin America — a mixed-race land where the curious are encouraged to follow "El camino del corazón." The path of the heart.
When the intellectual part of me pushed toward stability — education, employment — the savage heart pulled toward the horizon. There is more than one way to survive. Even for a systems engineer. It was, and still is, a path of solitude. Where the company of other seekers occasionally nourishes you. Or of others veering off course. Or of desolate beaches, canyons, rocks, and trees.
It should be said that a journey through wilderness, or one guided by the stars, is not the only path to fade the masks we don, or those the world places upon us. However, this experience of the primordial forces that shaped our planet can be uniquely insightful.
The above is not a criticism of Lévi-Strauss; the book was of a time when colonies were being "liberated" and debate about the nature of man was rife. Of course, we appear, to an extent, to have returned to those times — as we may have never really left them.
The savage in us doesn't disappear when we suppress it — it resurfaces, often in odd manifestations. Some in the bitcoin community extol an all-meat diet as a return to some ancestral state; India debates the moral status of "pure veg." These are minor attempts to channel, satiate or alienate the beast. But the same impulse, unchecked, produces sinister results that a significant proportion of humans would find difficult to digest. Livestreamed g______ is one such example — although it has begun to make its way through many a bowel.
The irony is that savage often connotes or is used to describe violence that is neither primitive nor unsystematic. Indeed, it is bureaucratised, institutionalised and optimised. Of course, the savage also murders and beheads. Evolutionary anthropologists examine the acts of violence tribes have and still do inflict on each other. What’s the difference? Domination and death are now engineered from a distance and at scale.
One common explanation is that this is simply because human societies have grown more complex than hunter-gatherer ones. But that's not quite right, either. As one researcher observes:
"When people talk about 'complex societies' or 'social complexity' it's often in relation to (...) population size or economic specialization, but many hunter-gatherer & other small-scale societies have very complex social systems regulating kinship, marriage & group identity"
That's the savage mind. And the engineer's mind. All vying in our still finite bodies to make sense of a species that emerged from the elements on a planet revolving around the sun in a vast universe.
That's the past and the present. The future now brings us a third one. An artificial — or rather, machine — mind.
One cross-section of humanity seeks to turn all human knowledge into a machine that takes us closer to understanding the unknown or that takes our place.
Here we return again to The Savage Mind, and another aspect of the book: decoding the mind from what one might now describe as first principles.
Faced with the apparent chaos that humans encounter, we observe differences. Differences enter into relations. Relations permit classification. Classifications become systems — a grid, a map, or a network through which the world becomes first intelligible and then actionable.
differences ↔ relations ↔ classification ↔ grid, map, network or system ↔ intelligibility or meaning ⇠⇢ application
This subsequently led us, and continues to lead us, to creating the world in a curious dialogue — or dance — with the rest of nature.
And how do large language models (LLMs) — the engines built into broader AI systems and agents — work?
tokens or features ↔ contextual relations ↔ representations ↔ distributed activations ↔ probable continuation ↔ human-read meaning ⇠⇢ application
The analogy should not be mistaken for an identity. A human classificatory system is not a neural network, and a neural network is not a human mind.
Both chains, however they're built, also point outward — toward modification of the social, material, and digital world. Rituals, tools, systems on one side; products, decisions, agents on the other. What differs isn't whether meaning matters, but how it's arrived at — one from lived, inherited constraint; the other, so far, from inference.
The machine mind is, for now, all engineer and no bricoleur — no wandering, no dead ends, no delight in the found object. Will the future be a child of both or the death of one?