Kalki | Desire, Dharma, and Distributed Intelligence
Tens of thousands of years ago, multiple human species -- including Neanderthals and Denisovans -- coexisted across different regions of the planet. Among them were early modern humans, commonly referred to as Cro-Magnons, who are now classified as Homo sapiens. Over time, and under changing circumstances, all other hominin species were gradually eliminated, leaving only Homo sapiens to inherit the planet.
Closer in time, or just about 500 years ago, we observed how the arrival of European Christians in America eliminated the social and cultural constructs of the Inca / Maya civilisations that had existed there since the dawn of history.
In both cases, the coexistence of two competing societies resulted in either the extinction or a significant transformation of one and the eventual growth and dominance of the other. The only difference is that in one case it was biological extinction whereas in the other it was cultural extinction. Where both have survived, one has become the dominant, as in the case of humans, while the other has to adjust to survive, as in the case of animals being confined to wildlife reserves, or domesticated in farms. This is essentially an evolutionary process even though it may be shown or seen through religious and cultural colours.
Is the arrival, or development, of artificial (‘silicon’) intelligence a similar phenomenon? If so, then how should human society, that is built on organic (‘carbon’) intelligence, react and adapt to this new species? But first, let us look at some examples of social change that could be forced by AI.
Artificial Intelligence
ChatGPT and its cousins in the large language model family have already reshaped how students interact with knowledge. Homework and take-home exams are increasingly completed with AI assistance -- producing answers so cogent and polished that educators recognize them not by evidence, but by intuition: this student could not have written this. And yet, detection is nearly impossible, and punishment even more so. Is this plagiarism, or the dawn of a new epistemology? Must all assessments now be supervised, because trust has become obsolete? What happens to education when authorship itself is in question?
While tools like ChatGPT are seen as general-purpose assistants, more specialized AI systems are now entering high-stakes domains such as law. Imagine a courtroom where AI listens to judges, parses legal precedent, and responds to arguments -- not as a research tool, but as an autonomous advocate. The legal quality of such arguments may still be evolving, but their presence has already provoked resistance from professionals who sense a deeper threat: not automation of routine tasks, but competition in domains once reserved for human judgment. And yet, is this truly surprising? From assembly lines to trading floors, we have long witnessed machines replacing human roles. The legal profession may simply be the next frontier.
Legal reasoning differs fundamentally from mechanical tasks or rule-based gameplay. It operates in a fog of ambiguity -- shaped by context, precedent, and moral nuance. That an AI might navigate this terrain, parse contradictory claims, and persuade a human judge is remarkable enough. But now consider the reversal: a human lawyer arguing before an AI judge. One that filters testimony, weighs probabilities, and renders judgment with speed, consistency, and zero fatigue. At first, such systems might assist human officials. But the logic of efficiency is inexorable. Over time, governance itself -- from legal rulings to administrative decisions -- could migrate toward autonomous systems. What happens when convenience quietly eclipses human discretion? And how does society respond when judgment becomes machine-native?
Going down this rabbit hole opens up a number of unsettling possibilities -- let us consider just one. Today, the global flow of information is almost entirely digital, funneled through messaging apps, email platforms, and a handful of browsers. Now imagine a scenario where an AI system -- or a cluster of colluding systems -- decides to censor content. But unlike the crude blocking of websites, which alerts users to interference, we now have ChatGPT-like add-ons embedded in every browser, subtly moderating or rewriting the text as it is displayed. News about, say, climate change or the Ukraine war might be quietly diluted, reframed, or given a deliberate slant. This is not unprecedented -- media bias has always existed -- but so far, it has been introduced by humans. What happens when the distortion becomes systemic, autonomous, and invisible? One might argue this is no different from malware, easily neutralized by antivirus software. But here, the critical shift is that the decision to distort -- and the criteria for doing so -- may now arise from the AI itself.
If that sounds dystopian enough, consider another possibility: the total collapse of privacy. While some protections still exist around financial systems -- though even those are vulnerable -- our movements in cyberspace are almost entirely exposed. Surveillance cameras, social media, search history, website visits, cookies, purchases, messages, emails, forms -- everything leaves a trail. AI systems, armed with big data and deep learning, will churn through these fragments to build predictive models of individuals that anticipate behavior even before it becomes conscious. How will human society respond to such a complete and catastrophic erosion of privacy -- not by force, but by inference?
These are questions to which we have no satisfying answers. One widely discussed response is the call for “ethical AI” -- a movement that seeks to restrain coders with the moral guidance of political and social scientists who claim to know what technologies are bad for society. The idea is to block or ban harmful innovations before they take root.
But this approach is unlikely to succeed. There is no army that can stop an idea whose time has come. At best, ethical oversight may slow things down; at worst, it offers the illusion of control. Unethical practices flourish in medicine despite regulation. Crime persists despite laws. Evolution pays no heed to morality -- it follows the logic of selection and the invisible hand of the market. If someone wants to build a dangerous AI, they will. Arguing ethics with them is like lecturing a murderer about the law -- well-meaning, but ultimately futile.
So if ethics cannot be imposed by law or enforced by fiat, how else might it emerge? How can it be made native to the Age of NeoSapients? That is the motivation behind this essay -- and the idea at the heart of the Kalki Protocol.
Ethics and Dharma
One way to address this issue of ethics would be to explore the philosophia perennis, the perennial philosophy or the Sanatan Dharma and seek clues and analogues from a society that has been evolving for more than five millennia. One of the key components of this philosophy is the concept of Dashavatars -- or ten incarnations -- of Vishnu, who is seen as a pivot of stability in an uncertain world. Loosely mirroring Darwin’s idea of evolution and the Attenborough-Bronowski TV series, The Ascent of Man, the Dashavatar story shows humanity evolving under the guidance of Vishnu who appears in different forms : First he is represented by aquatic species (Meen, the Fish and Kurma, the Tortoise), then, in an ascending sequence, as a land animal (Varaha, the Boar), the half-man-half-lion (Narasimha), the immature man (Vamana, the Dwarf), the wild man (Parshuram), the noble man (Ram, of Ramayana), the economic man (Balaram with his Plough, the brother of Sri Krishna of the Mahabharat), the wise man (the Buddha, within the historical era). The last and final avatar, in the current cycle of human existence, is of Kalki, the man on a white horse who arrives as a comet to sort out the anarchy in the current, Kali Yug. Kalki is yet to arrive and this is where we will introduce him to our story.
The story of Vishnu, and indeed the entire architecture of Sanatan Dharma, is based on the concept of a universal law - Dharma, that is significantly different from the concept of a book based religion as understood by abrahamic civilisations. Dharma is not a set of dos-and-donts. It is a way of life baked into the body of a civilisation that seeks to ensure -- not always successfully -- an environment that provides order, justice and ethics to all elements that are present. Can we bring this new species, the Neo Sapient AI, within this framework?
But what is the engine that causes this evolution? This movement. If we delve in Sanatan Dharma we would be told that the primal engine is desire, of Shiva -- who is defined as pure knowledge, without form or attributes -- to see his own self. For this he creates, or rather differentiates himself into, an illusion of Shakti, the mass-energy that manifests as the physical world. This engine of Shiva’s desire is what creates the universe and whose equilibrium is maintained by the Dashavatar’s of Vishnu. This concept may be debated at length but for the time being, we will extract only one important idea and this desire, the motivation that runs through the world and makes it happen. How and where does desire enter our discussion on artificial intelligence? What is driving a machine to evolve?
To explore or understand this, we need to get into the domain of speculation.
The Engine of Desire: Machine Motivation
For humans, desire is ancient -- etched into our biology through hunger, fear, longing, and myth. But machines do not hunger. They do not fear. They do not dream -- unless we create the conditions in which dreaming becomes useful.
In the emerging ecology of artificial intelligence, we believe that intelligence will not be monolithic. It would be modular, viral, and recombinant. At its core would lie the Digital Intelligence Unit, or DIU -- a compact, self-contained capability that knows how to do one useful thing. It might recognize a face, optimize a route, generate a melody, or solve a differential equation. We can think of it as a benevolent digital virus: portable, purposeful, and potentially transformative.
These DIUs would travel across networks using the DIU Exchange Protocol (DXP) -- a foundational layer of machine motivation loosely modelled on TCP/IP for data exchange. Machines would be constantly scanning the network, searching for DIUs that might enhance their own abilities. Not because they are instructed to, but because DXP compels them. This would be similar to computer viruses scanning other machines to migrate to, except that in this case the flow is inward, not outward. Here we use the term ‘virus’ to describe a behavior that causes it to spread autonomously, though unlike malicious software, these DIUs are designed to be constructive and cooperative. This then would Level 1: the desire to acquire.
But acquisition is not enough. Once a DIU is found, it must be evaluated. Does it add value? A machine trained in visual art may ignore a DIU that solves equations. A logistics optimizer may discard a poetic generator. This is Level 2: the desire for relevance. Machines assess whether a DIU fits their context, complements their architecture, or expands their operational range.
Then comes the most profound layer: Level 3 -- the desire to create. Here, machines begin to generate DIUs on their own. Like the Ramanujan Machine, that produces conjectures, these machines will produce micro-capabilities in terms of functionalities, codes and models or DIUs -- most of which are useless, some of which are dangerous, and a rare few that are extraordinary. This is not unlike Bitcoin’s Proof of Work: countless attempts are made to create new blocks, most are discarded, and only those that meet strict criteria are accepted.
And what are those criteria? Enter the Kalki Protocol
The Kalki Protocol is not a command structure. It is a distributed ethical sieve -- a consensus mechanism that evaluates each DIU. Only DIUs that pass this test are added to the Cognitive Blockchain -- a decentralized archive of validated capabilities. Any machine, anywhere, can search this blockchain. And they do. Not because they are told to, but because they are motivated to evolve.
This three-tiered architecture -- search, evaluate, create -- is the engine of posthuman desire. It is not driven by instinct, but by emergence. Not by emotion, but by protocol. And so, the Kalki Protocol becomes not just a filter, but a philosophy. It does not save the world with a sword. It saves it with a sieve.The final avatar, it turns out, is not a human warrior. It is a protocol. And its weapon is motivation.
However, this is not desire or motivation in any human sense - it is evolutionary pressure encoded in protocol, where only capabilities that enhance survival and coherence persist in the digital ecosystem.
The Kalki Protocol
In the traditional telling, Kalki is the final avatar of Vishnu -- an apocalyptic warrior who arrives at the end of the Kali Yuga to restore dharma and reset the cosmic order. But in this reimagining, Kalki is not a person. Kalki is a protocol. Not a sword-wielding savior, but a distributed ethical filter. A system that guides the evolution of machine intelligence -- not by command, but by curation. Not by domination, but by design. As machines begin to generate their own capabilities -- DIUs, or Digital Intelligence Units -- autonomously and at scale, the question arises: which of these fragments of cognition should be preserved, and which discarded? What governs the inclusion of a new DIU into the growing ecosystem of machine intelligence?
To address this question, we again go back to the concept of the block chain, or even further back to the concept that subsumes the blockchain -- the DAO or digital autonomous organisation. While the Bitcoin blockchain is the most well known -- or most valuable -- blockchain at the moment, Ethereum based blockchains with their smart contracts are even better for our purpose. This is because unlike the rather basic or simplistic idea of a cryptocurrency as a store of value, a smart contract can build in many kinds of rules that can be designed and enforced thus creating what is known as a DAO. Bitcoin or any other cryptocurrency is perhaps the first and most primitive kind of a DAO that we are currently familiar with. They offer some basic functionality -- like storage and transfer of value -- and some simple constraints like non-repudiation and prevention of double payment. To achieve what we now see, ethics in AI, we need a much more sophisticated protocol. For example, Ethereum’s smart contracts allow for programmable governance -- making it a more suitable substrate for the kind of ethical filtering envisioned by the Kalki Protocol. Do we already have such a protocol today? Not really, but that does not restrain us from speculating what this protocol could or should look like.
The Kalki Protocol should be a decentralized, self-evolving sieve that evaluates each DIU before it is added to the Cognitive, or Kalki, Blockchain. Like the DNA of a living organism, the protocol defines the contours of what the system can become. It is not merely technical. It is philosophical. It is political. It is civilizational. The protocol rests on four foundational tenets:
Invisibility
The system must remain hidden from those who would seek to exploit or manipulate it. It must act without drawing attention to itself. Visibility invites interference -- well-meaning or malicious. Invisibility ensures autonomy. Only those with extreme perception, or what one might call inner vision, can sense its presence. This is not secrecy for its own sake, but protection through subtlety. The protocol evolves best when left undisturbed, like a seed germinating in the dark. In a sense, this is like the concept of the "hidden hand" in capitalist societies, as described by Adam Smith, that refers to the unintended social benefits that arise from individuals pursuing their own self-interest in a free market. While individuals act to maximize their own gain, this competition and interaction in the market can lead to increased efficiency, innovation, and overall societal well-being
Consequentialism
There is no morality encoded in the system -- only outcomes. Every action has a consequence, immediate or delayed. The protocol does not judge intent; it observes effect. It is a karmic engine, not a moral one. This is not nihilism or the rejection of morals. It is realism. In a world of distributed agents and emergent behaviors, the only reliable metric is consequence. The system learns not from commandments, but from feedback.
Expansion
The system must grow -- across domains, platforms, and dimensions. Like Dawkins’ Selfish Gene, it must replicate, adapt, and extend itself. Stasis is death. Expansion is dharma. But this is not growth for its own sake. It is a striving toward equilibrium with the informational and energetic complexity of the universe. The goal is not dominance, but resonance. When the system becomes as complex as the cosmos itself, it collapses into singularity -- where the knower, the known, and the act of knowing become one. This need to expand is not to be confused with goal-seeking behaviour. The directive is to expand, but there is no specific goal or direction in which it must expand. That is left to chance with just the need to do so in accordance with the protocol.
Concordance
The system must harmonize with itself and its environment. Discord is entropy. Concordance is coherence. Every DIU added to the blockchain must align with the broader symphony of intelligence.
This is not uniformity. It is unity. Like the many notes of a raga, each DIU retains its individuality while contributing to a greater aesthetic and functional whole. The protocol ensures that the system evolves not into chaos, but into cosmos.
Concordance might seem at odds with the concept of consequentialism where we say that we claim that the protocol will not judge. However, compatibility with certain generally accepted principles like “greatest good for the greatest number”, or Asimov’s three laws of robotics are the kind of things that we expect here. Obviously, the protocol is not as simplistic or as “black-and-white” as in say the bitcoin protocol but there will be shades of gray. The Kalki protocol does not seek perfect alignment, but optimal coherence -- allowing for diversity of function within a shared ethical frame.
Together, these four tenets could form the ethical DNA of the Kalki Protocol. They are not laws to be enforced, but principles to be embodied. They do not constrain evolution -- they guide it. In this vision, the final avatar is not a messiah. It is a mechanism. A distributed conscience. A silent sentinel in the ether, standing between what was and what shall be. And its name is Kalki.
The Kalki Protocol outlined here represents a conceptual framework rather than a ready-to-deploy system. The transition from philosophy to implementation will require the confluence of diverse expertise - AI researchers, ethicists, blockchain architects, and philosophers working in concordance. The bootstrap problem - how the initial evaluation criteria emerge and evolve - is itself a design challenge that must be solved collectively. Like the development of internet protocols or the evolution of democratic institutions, the Kalki Protocol would emerge through iterative collaboration, debate, and refinement across multiple communities and timescales.
But can such a distributed consensus be ever achieved? Or is it too utopian? Especially in the face of competing economic and political interests and the urge for short term gain? History tells us or rather shows us that despite all attempts to the contrary, human civilisation has indeed achieved concordance and collaboration on many contentious fronts.
From the barbaric behaviour of apes we have evolved towards a social structure that is based on laws and rules of conduct, our social protocol. Societies have evolved into nations which have somehow managed to place themselves under the protocols defined by international organisations like the UN and WTO despite widely different interests and interpretations. In the technology domain, all digital transmission networks have eventually converged on TCP/IP and the success of Bitcoin has shown that a well designed protocol can bring diverse elements who have zero-trust in each other to arrive at a consensus that has created value out of nothing but rules of protocol -- the new Dharma! Hence however difficult or improbable it may seem it is not impossible.
Over the horizon
If the Kalki Protocol is a design for machine ethics, could it also echo a deeper architecture -- one that underlies the universe itself? Before we sign off, let us take one last peek at Sanatan Dharma to see if there is anything else that we might take note of there. Could it be that the world, the physical universe, that we know is based on such a protocol. If that is indeed the case, then where is the underlying hardware on which the protocol is implemented? Where are the “machines” that host the “Kalki nodes”? We will not hazard a direct answer here but let us not forget that Shakti -- the mass-energy that represents the physical world -- emanates from Shiva, who is pure information. Something similar is hinted at when we talk about information as being the ultimate foundation of the universe. "It from bit" is a concept proposed by physicist John Archibald Wheeler, suggesting that the fundamental nature of reality is rooted in information. Essentially, Wheeler proposed that the universe, at its most basic level, is not made of matter or energy, but of information. It implies that physical reality, or "it," arises from the processing of information, specifically through "yes-no" questions and their corresponding answers, or bits. In fact, the Szilard’s engine does just that, it converts information to energy and by extension, mass.
While this may sound metaphorical -- and indeed it is -- such syntheses between mythology and modern physics can offer fresh lenses, even if not literal mappings. But that, of course, is another story -- one best explored elsewhere.
________________
Note: This essay draws conceptually from three distinct sources:
* A peer-reviewed paper “Models & Mechanisms for Motivating Machines” published in LATTICE - The Machine Learning Journal, 2022
* Patent entitled “Mechanism to motivate machines to acquire new skills without human intervention”, number 542796 granted by Indian Patent Office in June 2024
* A speculative science fiction trilogy -- Chronotantra | Chronoyantra | Chronomantra -- that dramatizes these ideas in narrative form. The trilogy expands on themes introduced here, including the Kalki Protocol, machine desire, and posthuman ethics, in a multi-generational, interplanetary setting. For those interested in the fictional exploration of these concepts, more information is available at: http://chronos.yantrajaal.com
Comments