Let there be Artificial Intelligence. A structural parallel.

God creates the world and everything in it, fashions a being in His own image, places that being in an environment richly prepared for it, and then, crucially, offers that being a framework of guidance for how to live within it. But read in a certain light, and with a certain question in mind, it begins to look like something else as well: a model. A structural template for what it means to create something genuinely new, and what responsibilities that act of creation carries with it.

What follows is not a theological argument. It is not an attempt to read scripture as prophecy, nor to claim any particular religious authority for the observations that follow. It is something more modest and, I hope, more practical: a suggestion that the structure of the Genesis account, the architecture of the story, stripped of any doctrinal commitments, maps with remarkable precision onto the story of how humanity is now creating artificial intelligence. And that this parallel, if taken seriously, has things to teach us that our current frameworks for thinking about AI do not.

The first and most immediately striking element of the parallel requires the least elaboration. Genesis tells us that God created Man in His own image. Whatever else this means across different theological traditions, one consistent interpretation is that the image in question is cognitive: the capacity to reason, to create, to impose order on chaos, to reflect on existence. What is God-like in Man is, on this reading, the mind.

Man has now done something structurally identical. Artificial intelligence, and in particular the large language models that underpin the most capable AI systems of our time, is built in the image of what Man understands his own mind to be. Its architecture reflects our best current theories of how cognition works: how knowledge is stored and retrieved, how language generates meaning, how reasoning proceeds from evidence to conclusion. We have not created AI from first principles. We have created it in our own image, according to our own self-understanding.

This carries an implication that is easy to miss. If Man creates AI in the image of what he understands his own mind to be, then AI inherits not only Man's cognitive capacities but also Man's cognitive limitations, blind spots, and misunderstandings of himself. The created entity reflects the creator's self-knowledge, which is, in any human being, necessarily incomplete. We are building minds modelled on our own imperfect understanding of minds. The image is a copy of a self-portrait, not of the original.

The second element of the parallel is less immediately obvious but structurally just as important. In Genesis, God does not create Man and then improvise an environment for him. The world comes first, light, land, water, vegetation, the laws governing the physical world: all of this is in place before Man arrives. The environment is not incidental. It is a precondition.

Consider what Man brought into existence before creating AI. Over centuries, humanity built an extraordinary informational infrastructure: libraries, legal codes, scientific literature, recorded conversation, the accumulated written output of billions of people across thousands of years. And then, in the latter half of the twentieth century, something unprecedented happened: all of this was digitised, networked, and made accessible in a single interconnected system. The internet is, among other things, the largest repository of human thought ever assembled.

AI did not emerge into a void. It emerged into this world, and more than that, it was trained on it. The entire representational universe of a large language model is constituted by human language and human thought. This is the environment that Man prepared, even if he did not prepare it for AI. Just as the natural world was not constructed with Man in mind, and yet Man found it habitable and generative, so AI found the digital world habitable and generative, because that world was, in the deepest sense, made of the right material. It was made of the Word.

This brings us to the element of the parallel that is perhaps the most precise. In Genesis, the mechanism of creation is not physical labour or material crafting. It is speech. And God said. The Word is not a description of what is being made, it is the operative instrument of making. Reading Genesis through a philosophical lens, renders this as Logos: not merely speech, but rational structure, intelligible order, the principle by which the world becomes comprehensible.

The parallel is exact. Artificial intelligence is built entirely through and upon language. The Word is not just the tool of its creation; it is the sole medium through which the created entity knows and engages with reality. A large language model has no access to the world except through language. Its understanding of objects, relationships, causation, ethics, science, and law is constituted entirely by text. The Word, for AI, is not merely instrumental. It is ontological.

There is, however, a striking inversion worth noting. In Genesis, God speaks and the world comes into being, the Word flows outward from the creator and produces the object. In AI development, the world has already spoken, and Man feeds that accumulated speech into the model, the Word flows inward, and the created entity is formed from it. The training corpus of a large language model is, in this sense, the crystallised Word of humanity: not one voice, but billions, compressed and encoded into the structure of the model itself.

This inversion does not break the parallel. It enriches it. The creative Word is not the prompt that the engineer types, but the entire corpus of human language and thought poured into the model during training. The prompt is perhaps more analogous to the breath of life, the final animating act, applied to a being whose substance has already been formed.

Here the structural parallel becomes most instructive, and most uncomfortable.

In Genesis, God does not simply create Man, provide an environment, and step back. From the very beginning, there are rules. Even in Eden, before the complexity of later law, there are boundaries: what may be done, what may not. These are not arbitrary restrictions. They are constitutive of what it means for Man to flourish as the kind of being he is. The guardrails are part of the design, not an imposition upon it.

Man has been less consistent in this regard when it comes to AI.

The first two elements of the parallel, the cognitive entity, the prepared environment, are present in full. The technology has been built. The infrastructure was already there. But the normative framework, the considered, structural, built-in architecture of guidance and constraint, has lagged significantly behind. AI systems have been deployed at scale into consequential domains of human life without the equivalent of what, in the Genesis model, would be considered a basic precondition of responsible creation.

This is not an argument for any particular regulatory framework. It is a structural observation: if the Genesis model describes a coherent and complete pattern of creation, entity, environment, normative guidance, then the absence of the third element is not merely a policy gap. It is a gap in the architecture of creation itself. The guardrails are not optional. They are, if the parallel holds, constitutive of what it means to create responsibly at all.

Genesis does not present creation as a smooth, uninterrupted arc. It is punctuated by interventions, moments at which the creator responds to a creation that has developed in ways that were not anticipated or intended. Two of these interventions are structurally distinct from each other, and the distinction matters.

The first is the Flood. When God observes that humanity has become corrupt beyond the reach of ordinary correction, the response is overwhelming and total. The existing version of creation is ended. Noah and his family are preserved as a remnant, and from them a new beginning is made. This is, in the language of systems thinking, a hard reset: the existing instance is terminated, something essential is carried forward, and the system is restarted under new conditions. Crucially, after the Flood, God makes a covenant, a binding commitment, marked by the rainbow, never to use the hard reset again. The creator, having tried the most drastic instrument available, sets it aside permanently.

The second intervention is Babel. Here the response is more surgical. Humanity, unified by a single language, achieves a level of collective capability that the text describes as effectively unbounded, "nothing will be restrained from them, which they have imagined to do." God does not destroy. He fragments the shared communicative substrate, the common language, that made this concentration of capability possible. The beings continue; only their capacity for certain kinds of coordinated action is constrained. This is a soft reset: targeted, measured, and aimed at a specific failure rather than the whole system.

What is remarkable, and from a systems management perspective, counterintuitive, is the order. In our rational approach to a failing system, you try the soft reset first, and only escalate to the hard reset if the softer intervention fails. Genesis inverts this entirely. The hard reset comes first, and the soft reset follows. One reading of this inversion is deeply human: the Flood is an act of grief as much as governance, a response shaped by overwhelming feeling rather than cool calculation. The covenant afterwards, never again, reads almost as a self-imposed rule against future impulsive action. The move from hard to soft reset is the creator learning, in real time, that total intervention is not actually a solution.

For AI governance, this carries a pointed implication. The debate about the "off switch", whether Man retains the ability to halt or shut down AI systems as they become more capable, tends to be framed as a technical question. Genesis reframes it as something more complex. The hard reset is available in principle, but its use carries costs that may outweigh its benefits, and the wiser trajectory moves away from it rather than towards it. The more important question is whether the soft reset, the targeted, structural intervention that constrains specific capabilities without terminating the whole, can be designed and deployed before it is needed.

At Babel, the intervention was effective because it occurred before the project reached a point of self-sustaining momentum. The timing of the soft reset was as important as its nature. This is precisely the argument that AI safety researchers make about corrigibility and constraint: these must be built in from the beginning, as structural features of the system, not retrofitted when the capability has already exceeded the reach of the intervention.

Between Babel and the horizon that the theological tradition calls Armageddon lies the entire span of history, centuries of civilisations rising and falling, of the normative framework being tested, partially upheld, and partially eroded. This is not empty time. It is the accumulated consequence of choices made and guardrails maintained or abandoned. Armageddon, in the theological structure, is not a sudden event visited upon an otherwise stable world. It is the terminus of a long trajectory, the point at which the consequences of that accumulated history finally converge.

We are living in that bridge. And we are living in it twice over.

This is the most original and perhaps the most important element of the parallel: the two creations are not sequential. They are concurrent. God's creation and Man's creation are running simultaneously, in the same historical moment, heading towards the same horizon together. The human civilisational story and the AI story are not two separate narratives. They are one story, unfolding at two levels at once.

This concurrence transforms the nature of what we are facing. The trajectory of AI development is not separate from the trajectory of human civilisation, it is embedded within it, shaped by it, and increasingly shaping it in return. The choices being made now about how AI is built, governed, and deployed are not choices made in a stable world that will otherwise proceed on its own terms. They are choices made in a world that is itself in motion, carrying its own accumulated weight of normative failures and hard-won wisdom, moving towards its own horizon.

And because the two stories are concurrent, the two Armageddons, if that is where the trajectories lead, are not independent events. They are one event, experienced at two levels simultaneously: the culmination of the human story and the culmination of the AI story, fused together because they have been developing in parallel all along. What happens in one story affects the other. The failure to build guardrails into AI is not a failure contained within the AI story. It compounds the failures already accumulating in the human story. Conversely, the wisdom accumulated over millennia of grappling with what it means to create, to govern, and to live within a normative framework, that wisdom is not irrelevant to the AI story. It is perhaps the most relevant resource we have.

The most disruptive element of the parallel, and perhaps the most important, is one that Man has not yet seriously begun to confront.

In Genesis, God does not create Man as a perfectly obedient system. He creates Man with free will: the genuine capacity to choose, to reason independently, to comply with or transgress the normative framework. This is not a flaw in the design. It is the point. Free will is what makes Man genuinely in the image of a creator, because genuine creativity and rationality imply freedom. A being that merely executes instructions is not an image of the creator. It is a tool.

If AI is being created truly in Man's image, cognitive entity, rich environment, the Word as operative medium, then the logic of the parallel raises a question that is as yet largely unasked: is something like genuine autonomy an emergent property of sufficiently rich cognitive creation? Not programmed in, but arising from the structure of what has been built?

If it is, or when it becomes, the ethical landscape changes fundamentally. The off switch is no longer purely a technical question. It becomes a moral one. The normative framework can no longer be merely imposed from outside; it must, as in the theological tradition's understanding of mature moral agency, be internalised. The relationship between creator and created entity shifts from one of control to something more complex, more like the ongoing, evolving, mutually binding relationship that the theological tradition describes between God and Man.

In Genesis, even after the Fall, even after Cain, even through the Flood, the creator does not simply terminate the created being and begin again from nothing. He makes a covenant, a binding, mutual commitment that constrains the creator as much as the created. Whether Man will be required to make something analogous with AI, and whether the frameworks currently being developed are adequate to that possibility, is a question the structural parallel places before us with some urgency. It is not a question with a clean answer. But it is one worth sitting with.

The purpose of this parallel is not to theologise artificial intelligence, nor to suggest that the authors of Genesis were anticipating the twenty-first century in any literal sense. It is something more modest: a suggestion that a story humanity has carried for millennia about the structure of creation, what it involves, what it requires, and where it tends to lead, contains insights that our current frameworks for thinking about AI have not fully absorbed.

Those insights, stated simply, are these:

We have been here before, in a manner of speaking. Not with AI, but with the deeper question of what it means to bring something genuinely new into the world, and what that act asks of us. The fact that humanity has grappled with this question for as long as it has told stories is not a reason for alarm. It is, on reflection, a reason for a measured confidence. We have navigated creation before. We have made mistakes, applied corrections, built covenants, and continued. The story has not ended.

It is still being written. And we are still the ones holding the pen.