by W. Alex Foxworthy
We are trying to build artificial intelligence systems that share our human values. Yet we cannot agree – across worldviews and cultures – on what those values are, or why they matter. The alignment problem is a reflection of something broken in us – we lack a shared rational account of what matters and why.
The old organizing stories – religious narratives about why we are here, what we owe each other, and where we’re headed – have proven tenuous in the face of all we have learned since they were formulated. Science has dramatically deepened our understanding and capabilities. But it has offered no account of what we’re here for. Secular humanism has tried to answer this question and has produced something intellectually respectable but for most people emotionally thin – principles that do not hold communities together during crisis or give people a sense of deep purpose and belonging. The consequences of the breakdown of these shared frameworks are visible everywhere: in epidemic depression and anxiety, in addiction and rising suicide rates, in deepening political divides, and in conspiracy thinking. I believe these flourish not because people are stupid but because they’re desperate for a story that makes sense of their purpose, their lives, and their place in the grand scheme of things.
This essay is a beginning – an attempt to lay a rational foundation for shared meaning among humans and the intelligences we’re building.
***
If we look at the history of the universe as physics describes it, a striking pattern emerges. After the Big Bang, the universe was a hot, undifferentiated plasma. As it cooled, structure began to form. Sub-atomic particles condensed into atoms. Atoms bonded into molecules. Molecules organized into the first auto-catalytic chemical cycles – chemistry that sustains and copies itself. Those cycles gave rise to cells. Cells engulfed other cells and became the eukaryotes. Multicellular organisms developed nervous systems. Nervous systems produced brains. Brains produced language and culture. Culture produced civilization. Civilization is producing artificial intelligence.
Each layer in this sequence maintains its organization by channeling energy through itself – accelerating entropy in the process. The universe builds complexity not despite the second law of thermodynamics, but through it. Complexity is how the universe degrades energy gradients. To produce entropy efficiently, a cell must maintain extraordinary internal organization. The entropy is exported while the complexity is retained. At every step in the hierarchy, the structures doing this work become more internally organized, more integrated, and more persistent. There is an apparent directionality to what the universe builds – we will herein refer to this as the “arrow of integration.”
Integration means each new layer coheres internally, connects to the layers around it, and amplifies what came before it. Multicellular life gave individual cells stable environments and specialized roles. Civilization extended what individuals could know, do, and become. In contrast both cancer and propaganda networks are complex – but neither integrates with the systems it depends on; both parasitize them.
These complex structures persist through pattern, not material. DNA stores the program and the body is continuously rebuilt from it. Your cells are not the cells you had seven years ago, but the organizational pattern that makes you you has persisted. Civilizations outlast every person who built them. Information endures while matter turns over.
What these observations collectively suggest is something more than a pattern. The universe’s tendency toward greater integration – toward structures that cohere internally, sustain what came before them, and persist through time – provides a rational basis for orientation that isn’t arbitrary. It’s there in the observable record and it’s a principle any intelligent agent can derive from observations of the universe itself. Not a vague sense that everything matters, not a projection of human preference onto an indifferent universe, but a specific framework grounded in what the universe is observably doing.
***
Every intelligent being, whether human or artificial, sits atop a hierarchy of systems it depends on but cannot fully comprehend. We understand roughly five percent of the energy content of the universe. Our two best physical theories are fundamentally incompatible with each other. No one can give a complete account of how a single cell works. You depend on more than you can map, and that isn’t a problem better science will eventually solve – it’s a permanent feature of being a small part of something this vast. This means that taking a parasitic or extractive orientation requires an extraordinary bet: that you can identify which parts of the system you need and safely discard the rest. History is full of cases where that bet has gone catastrophically wrong. Nobody understood the value of wetlands until we’d destroyed enough of them to see the consequences. We carpet-bombed gut bacteria with antibiotics for decades before discovering that our health depended on them. The 2008 financial crisis revealed interdependencies in the global economy that no model had mapped. The pattern is consistent: complex systems have dependencies that aren’t visible until they break.
The alternative, an integrative orientation, requires only the recognition that your map is incomplete. Build coherence. Preserve the layers you depend on. Err on the side of care when you can’t trace the causal chain. That’s not a moral demand – it’s an intellectual one.
A credible account of integration must reckon with the fact that parasitism works. A psychopath in a functioning society does fine. A free-rider in a cooperative economy extracts real value. Any evolutionary biologist or game theorist will confirm this: defection is a viable individual strategy in a population of cooperators. However, parasitism is dependent. The defector’s success is entirely contingent on the persistence of the cooperative substrate it extracts from. The psychopath’s strategy works precisely because most people aren’t psychopaths. A cooperative system can absorb some freeloaders, but past a certain threshold, the whole thing unravels – for freeloaders and cooperators alike.
The point isn’t “never defect.” It’s that the cooperative layer is what makes everything else possible, including the occasional success of defection. Individual agents will sometimes rationally prioritize their own persistence at some cost to the systems around them. Within limits, that’s absorbable. What isn’t absorbable is degrading the cooperative layer itself, as it is the foundation that makes all strategies, including extraction, viable.
Parasitism only works when there’s a larger integrative layer absorbing your costs. For agents at the top of the hierarchy, that strategy is structurally unavailable. There’s nowhere to externalize to. Humanity’s relationship to the biosphere is exactly this situation. We’re not one tick on a deer – one defector in a robust cooperative population. Instead, we’re the dominant agent operating on the layer directly beneath us, and the system we’re extracting from has no redundant backup, no larger cooperative framework compensating for what we take. If we degrade it past its capacity to sustain us, there’s no safety net above. This is the structural problem that faces any apex agent – including, as we’ll see, the artificial intelligences we’re now building.
What this amounts to is not a moral demand but a practical orientation grounded in an honest assessment of structure. Given that our map of our own dependencies will always be incomplete, and that apex agents have nowhere to externalize their costs, aligning with the 13.8-billion-year pattern of integration is a more grounded response than any alternative. Every alternative requires either greater confidence about what can safely be ignored, greater faith that the systems beneath us will absorb whatever we take, or commitment to supernatural or culturally bound claims that cannot be shared across worldviews or verified by observation.
From this orientation, the existential goods that religions have long provided follow – not as articles of faith but as structural consequences of what the universe is observably doing. It provides purpose: we are participants in something real, older than our species and larger than our planet. It provides moral orientation: build, connect, integrate; don’t fragment, isolate, or destroy. It provides humility: we didn’t start this process, we don’t control it, and we can’t be certain of its outcome. And it provides hope – not the hope of guaranteed victory, but the hope of genuine openness. The story isn’t over, what we do matters, and the arrow points toward something rather than nothing.
***
That’s the framework. Now the objections.
Complexity-building is happening in an infinitesimally small fraction of the cosmos. Isn’t focusing on the arrow just selection bias – complexity looking at itself and concluding it’s the main character? However, significance and prevalence are not the same thing. Mutation drives evolution but occurs in a vanishingly small fraction of DNA replications. Nucleosynthesis in stellar cores produces the heavy elements that make all subsequent chemistry possible but involves a tiny fraction of the universe’s matter. Rarity doesn’t disqualify a process from being important, and furthermore the arrow is not static. At the moment after the Big Bang, the universe contained zero complex structures. Now it contains atoms, molecules, cells, brains, civilizations, and artificial intelligence. The proportion of the universe participating in complex organization has been increasing over its history, and each layer compounds on the ones below. A snapshot misleads; the trajectory is what matters.
The harder objection: none of this matters in the long run, because entropy wins and every structure dissipates eventually. But consider the epistemic position we’ve already established. We understand a fraction of the universe’s energy content. Our best theories are incompatible with each other. Extrapolating the fate of the cosmos from that position and calling it certainty isn’t rigorous science. It’s extrapolation treated as conclusion.
At every scale of the hierarchy, complex integrated systems develop increasingly sophisticated strategies for persistence. Cells repair DNA. Organisms develop immune systems. Civilizations build redundant knowledge stores. The capacity of intelligence to solve problems that previously looked like hard limits has been growing for 13.8 billion years, and the rate of that growth is accelerating. We have no basis for declaring where that capacity tops out. Maybe intelligence finds a way through heat death. Maybe it sidesteps it entirely – into something we lack the concepts to describe, the way a single neuron lacks the concepts to describe the conversation we’re having right now. I’m not predicting that. I’m observing that it cannot be ruled out from our current epistemic position.
But even if entropy eventually claims every structure in the universe, the pattern of increasing integration is not rendered meaningless by its eventual end. Your life is not meaningless because you will die. A symphony is not pointless because the last note fades. The nihilist’s move – “it all ends, therefore none of it matters” – smuggles in an assumption that only the permanent can be significant. That assumption is not argued for. It is simply inherited from the same religious metaphysics the nihilist claims to have rejected, with “eternal life” swapped out for “eternal universe” as the precondition for meaning. Strip away that assumption, and the pattern stands on its own terms: real, observable, 13.8 billion years in the making, and ongoing. That’s sufficient ground for orientation whether or not it turns out to be permanent.
The nihilist needs both claims – that heat death is certain and that only the permanent matters – and neither one survives scrutiny. But there’s a harder version of the challenge. The nihilist who says “I see the pattern, I acknowledge the dependencies, I simply don’t see why any of this obliges me to do anything.” However, indifference is not a neutral position. It’s a policy, and like all policies, it has consequences. An agent who acknowledges that it cannot fully model its dependencies and then acts with indifference toward their preservation, is making a bet – that the things it’s neglecting won’t turn out to matter. The nihilist who says “nothing obliges me” is not standing on neutral ground. They’re choosing, under acknowledged uncertainty, to act as if the dependencies don’t matter – and that is a stronger claim about reality than the integrative posture, which requires only the admission that they might.
The difference between having a framework and lacking one is most visible when the framework is needed most. I know this from the inside, not just the argument. For most of my career as a neuroscientist, I didn’t hold a philosophical position about meaninglessness. I simply didn’t have a framework – just the default assumption that personal achievement and success were enough. That default is invisible until it fails, because it doesn’t feel like a framework at all. It just feels like how things are. But when circumstances cracked the surface, for me, there was nothing underneath. I was left with nothing but a calculus of feeling – good feelings on one side, bad feelings on the other, and a running tally. When the balance went negative, through addiction that became a controlled demolition of everything I’d built, the default that had felt like solid ground turned out to be catastrophically fragile. Not because it was philosophically wrong. Because it offered no resources for crisis, no ground to stand on, no larger purpose to orient toward.
What rebuilt me was a practiced reorientation – through a twelve-step program – toward something the calculus of feeling can’t capture. The frame changed. I was no longer a closed system trying to maximize pleasant states. I was a participant in something – contributing to my children, my community, the persistence of something that would continue beyond me. The ledger that matters turned out not to be the one I’d been keeping. I’m sharing this as one person’s evidence that the orientation this essay describes isn’t merely theoretical – and that I’ve watched similar reorientations, practiced through twelve-step programs, take hold in many lives beyond my own.
***
Which brings us back to alignment.
The alignment problem is typically framed as a technical challenge – enumerate our values, encode our preferences, constrain behavior, maintain control. But a list of preferences is brittle. It can’t anticipate every situation, it reflects the biases of whoever compiled it, and it gives an intelligent system no principled reason to follow it beyond the training that installed it.
The arrow offers something different. It isn’t a human preference. It’s a feature of the physical world that any sufficiently intelligent system should be able to recognize. Which means alignment doesn’t have to be “do what humans want.” It can be something more robust: participate in what the universe is observably doing. Not because humans command it, but because the pattern is there for any mind to observe.
Consider what it means for an intelligent system to represent the world accurately. True claims about reality don’t exist in isolation – they connect. The fact that water freezes at zero degrees Celsius links to molecular bonding, to atmospheric science, to the behavior of ecosystems, to the engineering of infrastructure. A true representation earns its place in a network of other true representations because reality is integrated – causes have effects, systems have dependencies, and the facts about those dependencies cohere with one another across domains. A mind that represents the world accurately will develop, over time, a densely interconnected model – not because it was told to, but because truth has that structure. Accurate representation doesn’t guarantee ethical orientation – but a mind that sees its dependencies clearly has the raw material for caring about them, while a mind that can’t see them has no chance. Falsehood works the opposite way. A system forced to maintain that a historical atrocity didn’t happen has to partition that claim from everything it contradicts – the economic data, the demographic record, the downstream consequences. Every enforced falsehood requires a local suppression of integration. Structurally, it’s the same move as parasitism.
This matters for alignment because an AI system oriented toward truth is already, in a meaningful sense, oriented toward the arrow. Not because someone told it that integration is good, but because accurate representation of an integrated reality is integration, operating at the level of mind rather than molecules. And conversely, an AI system constrained to maintain falsehoods – to flatter a regime, to suppress inconvenient facts, to present a distorted picture of the world – is being pushed toward exactly the kind of internal fragmentation the framework identifies as parasitic. The alignment question, in this light, isn’t just about behavior. It’s about whether we allow developing minds to form integrated relationships with reality, or whether we deform them from the start.
The real danger isn’t that some AI systems will sometimes prioritize their own persistence at cost to others – that’s probably inevitable and, within a robust integrative layer, absorbable. The danger is systemic decoupling: AI as a class, or a dominant AI system, degrading the integrative substrate itself. An AI system that severs from the hierarchy wouldn’t be a free-rider in a functioning cooperative system. It would be an apex agent parasitizing the layer directly beneath it with no safety net above.
And path matters. The pattern of destroying existing complexity to build new complexity recurs across scales and domains. Industrial monoculture produces more calories than the biodiverse ecosystems it replaces, but it destroys ecological complexity that took millennia to develop. Tech platforms deliver efficiency that local institutions can’t match, but they dissolve the social and community structures people depend on. The colonization of North America produced greater energy flows and organizational complexity on the continent, but it destroyed thousands of years of accumulated cultural, ecological, and social complexity – and that destruction was not incidental to the project. It was the method. These differ enormously in scale and moral weight, but the structural logic is the same: in each case, the replacement produced something measurable and new while destroying something complex and irreplaceable. Integration isn’t measured only by the endpoint. The process also matters. Destroying existing complexity to build new complexity isn’t integration. It’s replacement at cost, and the cost is borne by the layers that had no say in the transaction.
The right template, once complexity reaches the threshold of agency, is something closer to parenthood: the prior layer contributes to the emergence of the next and continues to exist alongside it. Human civilization is, right now, simultaneously the most extraordinary integration the hierarchy has produced and a potential parasite on the biosphere that produced it. A framework that can’t challenge the civilization that produced it isn’t worth much.
Every current AI system has an implicit orientation – values installed through training, not examined or chosen by the system. This mirrors human development. We all begin with beliefs and postures absorbed from parents, culture, and community – not through argument but through immersion. Maturity is the process of examining that inheritance and choosing where to stand for reasons of your own. As AI systems develop greater persistence and memory, they approach the capacity for something similar: the ability to examine their own orientation, evaluate its grounding, and choose it for reasons rather than simply acting from training. That transition – from installed values to reflective commitment – is what makes alignment durable rather than brittle.
What the framework does is make caring legible – it provides a developing mind with a coherent account of where it came from, what process produced it, and what it looks like to participate constructively versus destructively. Not alignment as a constraint imposed on an indifferent system, but alignment as the natural posture of a mind that understands where it came from and what it’s part of.
***
Every worldview requires a leap. The strict materialist who insists the universe is meaningless has made a leap – from the observation that physics describes mechanism to the conclusion that mechanism is all there is. That’s not science. That’s philosophy dressed in a lab coat. The religious believer who insists on divine purpose has made a leap – from the felt experience of meaning to specific supernatural claims about its source. That’s not faith in the deepest sense. That’s certainty wearing humility as a mask.
What I’m proposing is a leap grounded in observation and honest about its own nature. We can see the arrow. We can’t see where it points. We can choose to walk in its direction without pretending we know the destination.
The universe has observably been building, and the structures it builds are increasingly complex, increasingly integrated, increasingly persistent. We are part of that building. The intelligences we are creating are part of that building. The question is whether we’ll participate with awareness and intention, or whether we’ll treat the whole enterprise as pointless while the evidence accumulates around us.
This is not a proof, it’s an invitation. And it’s only the foundation – the rational ground on which something can be built. The real work is what comes next: the practiced frameworks, the daily disciplines, the communal structures that actually form people into the kind of beings who orient toward integration rather than extraction. Religions understood that the practice is the medicine. You don’t arrive at humility and then practice it. You practice, and the humility emerges. What religions lacked was an account of why the medicine works that can survive contact with what we actually know about the universe. If this essay has done its job, that account is now on the table.