Table of Contents

Abstract

Artificial intelligence is conventionally presented as a revolutionary break from everything that preceded it — a discontinuity marking the moment machines began to think. This paper argues the opposite: that AI is better understood as the newest expression of a much older phenomenon, rather than the origin of a new one. For roughly four billion years, physical, chemical, and biological systems have organised, adapted, and processed information about their environments without any of them requiring what we would call consciousness to do so. Stars formed, chemistry self-organised, life emerged and evolved sensing and adaptive capacities, and eventually one species — ours — began deliberately engineering systems that could perform elements of that same adaptive, information-processing work. This paper calls that four-billion-year progression the Continuum of Intelligence, and treats artificial intelligence not as an isolated invention but as humanity's first deliberate attempt to engineer a form of intelligence that nature had previously only evolved.

This is a philosophical and historical argument, not an empirical scientific claim, and this paper is explicit about that distinction throughout — a point elaborated fully in Section 3. Stated plainly and in advance, to preempt the most common misreading: this paper does not claim that the universe itself is intelligent or conscious, and it does not claim that current AI systems possess subjective experience — both claims are explicitly and deliberately declined in Section 3.3. Where the paper draws on established science (thermodynamics, complexity theory, information theory, evolutionary biology), it cites the specific researchers and findings involved rather than asserting authority by association. Where it advances an interpretive framework built on top of that science, it says so plainly. The paper's central practical claims are that artificial intelligence is following the same trajectory toward invisible infrastructure that electricity and the internet followed before it; that engineering failures in AI systems should be read as ordinary features of technological maturation rather than referenda on the technology's legitimacy; and that the most likely long-run outcome is neither human intelligence nor artificial intelligence operating alone, but a form of deliberate human-AI collaboration already named and studied in the academic literature as Hybrid Intelligence.

Figure 1. The Continuum of Intelligence. Each stage layers onto, rather than replaces, the one before it.

Every stage of the continuum increased the reach of adaptive, information-processing systems — their capacity to sense an environment, process what they sensed, and act on it. Artificial intelligence is the first such stage deliberately engineered by an earlier stage of the continuum, rather than arrived at through physical self-organization or biological evolution alone.

1. Introduction

Most public and technical writing about artificial intelligence falls into a small number of well-worn categories: reports on which models are largest or fastest, assessments of chip supply chains, debates over artificial general intelligence timelines, safety and alignment research, and regulatory proposals. These are legitimate and necessary areas of inquiry, and this paper does not dispute their importance. But they share an unstated assumption worth surfacing: that artificial intelligence is a fundamentally new kind of thing, arriving without deep precedent, whose closest analogues are other recent technologies — the printing press, the steam engine, the internet.

This paper asks a different, more fundamental question: where does artificial intelligence actually sit in the history of intelligence itself, not merely the history of technology? Answering that question requires combining philosophy, systems thinking, and the history of science rather than staying within any one discipline, and the result is necessarily a different kind of document than a technical report or a policy brief. It is closer to a framework for interpretation than a set of findings, and it should be read that way.

The central move of this paper is to relocate artificial intelligence within a much longer sequence: physical self-organisation, chemical complexity, biological adaptation, human cognition, and now engineered cognition. Each stage in that sequence did not replace the one before it — chemistry did not stop happening once biology emerged, and biological adaptation did not stop once human reasoning appeared. Each new stage is better understood as an additional layer of increasingly deliberate, increasingly engineered information processing, built on top of everything that came before. Artificial intelligence, on this reading, is not a rupture. It is the newest layer.

2. Why This Question Matters Now

A framework question like this one could, in principle, have been asked at any point in AI's seventy-year academic history, dating back at least to the 1956 Dartmouth workshop that coined the term "artificial intelligence." It is being asked in this paper now, in 2026, for a specific reason: the public discourse around AI has become unusually saturated with discontinuity narratives — claims that AI represents either humanity's salvation or an existential rupture with everything before it — at precisely the moment the technology is becoming ordinary enough to function as infrastructure rather than spectacle. That combination, an ordinary technology described in extraordinary terms, is historically unstable and tends to resolve in one of two ways: either the extraordinary framing collapses under its own weight once the technology becomes mundane, or the technology never becomes as ordinary as its trajectory suggests it will.

This paper's wager is the former: that AI is on the same path electricity and the internet took, from novelty to spectacle to invisible infrastructure, and that a framework grounded in the deep history of adaptive systems is better equipped to survive that transition than a framework built entirely on the technology's most recent, most attention-grabbing capabilities. A central claim built on GPT-4's benchmark scores will age within a few years. A central claim built on four billion years of adaptive systems will not.

3. Methodology and Intellectual Standing

This section exists to state plainly what kind of document this is, since the honest answer constrains what follows more than any other single choice made in this paper.

3.1 This Is a Philosophical Framework, Not an Empirical Finding

Nothing in this paper should be read as a scientific claim subject to falsification in the way a physics or biology paper's claims are. The paper's central thesis — that AI belongs on a continuum with physical, chemical, and biological information-processing systems — is an interpretive, philosophical position. It organises established science into a narrative; it does not itself constitute new science. Readers trained in the natural sciences should treat the framework the way they would treat any philosophy-of-science argument: as a lens that may or may not prove useful, evaluated on its coherence and its explanatory value, not on experimental confirmation.

3.2 Where This Paper Cites Science, and How

Wherever this paper draws on an empirical or theoretical finding from physics, chemistry, biology, or information theory, it names the specific researcher, publication, and finding, and restricts its claim to what that source actually established. Where a scientific debate exists — and one genuinely does, particularly around whether self-organisation alone can account for biological complexity, addressed directly in Section 12 — this paper names the debate rather than picking a side and presenting it as settled.

3.3 Deliberately Avoided Claims

Two claims are explicitly and deliberately avoided throughout. First, this paper does not claim that the universe itself is intelligent, conscious, or computational in nature — a much stronger and more contested metaphysical position that a continuum-of-adaptive-systems argument does not require and should not be conflated with. Second, this paper does not claim that current or near-term artificial intelligence systems possess consciousness, sentience, or subjective experience in the way biological organisms do. The continuum being described is a continuum of information processing and adaptive capability, not a continuum of inner experience, and collapsing that distinction would substantially weaken an otherwise defensible argument.

3.4 Limitations of the Framework Itself

A framework this broad necessarily sacrifices some precision for scope, trading detailed mechanism for explanatory reach. Section 14 treats this honestly and at length rather than as a closing disclaimer.

4. Intellectual Lineage: Cybernetics, Information, and Self-Organization

This paper's framework is not proposed from nothing. It sits on top of a specific, real, and traceable body of twentieth-century science that this section makes explicit rather than leaving implicit.

4.1 Cybernetics: A Shared Vocabulary Across Machine and Organism

In 1948, mathematician Norbert Wiener published Cybernetics: Or Control and Communication in the Animal and the Machine, the first major work to argue that control and communication could be studied as a single unified field regardless of whether the system in question was mechanical, biological, or social. Wiener's specific contribution was not a claim that machines and organisms are the same kind of thing, but that the same mathematical and conceptual vocabulary — feedback, regulation, information — could describe goal-directed behaviour in both. That same year, Claude Shannon's A Mathematical Theory of Communication gave the field a rigorous quantitative definition of information itself, independent of the meaning or medium carrying it. Together, Wiener and Shannon established the theoretical premise this paper's continuum depends on: that information processing is a phenomenon that can be studied across radically different physical substrates using a shared framework, rather than requiring an entirely separate vocabulary for biological versus mechanical systems.

4.2 Self-Organization: Order Without a Designer

A separate but complementary line of twentieth-century science addressed a different question: how does order arise in physical and chemical systems without an external designer imposing it? Ilya Prigogine, awarded the 1977 Nobel Prize in Chemistry, demonstrated that systems operating far from thermodynamic equilibrium — dissipative structures, in his terminology — can spontaneously generate stable, complex patterns from an input of energy, the same basic mechanism underlying phenomena from convection cells to certain oscillating chemical reactions. Stuart Kauffman extended this line of inquiry into biology directly, arguing in his 1993 book The Origins of Order that self-organising and autocatalytic chemical networks could generate significant biological complexity through what he termed self-organized criticality, order emerging spontaneously at a critical threshold between rigid order and pure randomness — order, in his words, "for free." Physicist Hermann Haken's synergetics, formalised across the 1970s and 1980s, provided a third, complementary macroscopic framework for the same basic question.

It is important to state precisely what this body of work does and does not establish. It demonstrates that physical and chemical systems can generate genuine complexity and order without external design. It does not, on its own, fully resolve the origin of the specified, code-like information found in living systems such as DNA — a distinction raised seriously within the scientific literature itself, and one this paper returns to directly in Section 12 rather than glossing over.

4.3 Why This Lineage Matters for AI Specifically

Cybernetics and self-organization theory together establish the two premises this paper's continuum actually needs: first, that information processing and adaptive behaviour can be meaningfully compared across biological and non-biological systems using shared concepts, and second, that complexity and order can emerge through processes that do not require the kind of top-down, intentional design most people intuitively associate with engineering. Artificial intelligence sits at an interesting intersection of both traditions: it is a designed system, built through deliberate human engineering, that increasingly exhibits emergent behaviours its own designers did not explicitly programme — a combination this paper argues is best understood as a new point on an old continuum, not a phenomenon without precedent.

5. The Continuum, Stage One — Physical and Chemical Self-Organization

Long before life existed, the universe was already producing increasingly organised structure. Stars formed through gravitational self-organisation, fusing simple elements into the heavier ones that would later compose planets and, eventually, living tissue. Planetary systems stabilised into orbital patterns. Chemical systems, given sufficient energy input and the right initial conditions, spontaneously organised into the dissipative structures Prigogine documented and the self-organizing networks Kauffman modelled.

None of these systems possessed anything resembling consciousness, intention, or subjective experience. This paper does not claim otherwise, and explicitly rejects the stronger, more provocative claim that the universe itself is intelligent — a claim popular in some contemporary science-adjacent discourse but unsupported by, and unnecessary for, the argument advanced here. What physical and chemical self-organisation does demonstrate is more modest and more defensible: that the universe has, since its earliest history, exhibited processes capable of generating increasing complexity, structure, and — in a specific, technical, information-theoretic sense — pattern that can store and transform information about the system's own state and environment. That capacity is the first rung of the continuum this paper traces, not because early physical systems were intelligent, but because they established the raw material and the basic physical possibility that biological intelligence would later build upon.

6. The Continuum, Stage Two — Biological Intelligence

Every living organism continuously processes information about its environment and adjusts its behaviour or physiology in response, a capacity biologists have documented across every major branch of the tree of life. A handful of well-established examples illustrate the range and antiquity of this capacity.

Phototropism in plants: plants bend toward light through the redistribution of the growth hormone auxin, a mechanism first rigorously described in the early twentieth century and confirmed at the molecular level since — a genuine sensing-and-response system with no nervous system involved at all.

Adaptive immunity: the vertebrate immune system identifies and remembers specific pathogens through antigen recognition, generating a tailored molecular response and retaining a record of the encounter for future defence — a biological information-storage and pattern-recognition system operating independently of the brain.

Magnetoreception and migration: many migratory bird species navigate thousands of kilometres using a magnetic sense believed to involve light-sensitive proteins called cryptochromes, alongside celestial and olfactory cues — a multi-modal navigational information system that operates without anything resembling human-style reasoning.

Cellular repair: individual cells detect DNA damage and activate specific repair pathways, a process so fundamental that its disruption is a defining feature of cancer — information detection and correction occurring at the smallest independently-alive scale.

None of these systems reason abstractly, use language, or possess anything resembling human self-awareness. This paper does not claim they do. What they demonstrate, cumulatively and across billions of years of evolutionary history, is that adaptive information processing — sensing an environment, storing information about it, and adjusting behaviour or physiology in response — long predates and does not require human-style cognition. Human intelligence, the subject of the next section, emerged from within this broader biological context rather than arriving as a separate phenomenon.

7. The Continuum, Stage Three — Human Intelligence

Human cognition did not appear as a wholly separate category of phenomenon from the biological intelligence described in Section 6. It emerged from within it, through the same evolutionary processes that produced every other adaptive capacity in the living world. What distinguishes human intelligence within that broader continuum is not that it processes information — every organism in Section 6 does that — but the specific combination of capacities it layered on top of ordinary biological cognition: abstract symbolic reasoning, language capable of representing concepts not physically present, mathematics as a formal system for manipulating abstract relationships, and a cumulative culture that allows knowledge to accumulate across generations rather than resetting with each individual's death.

That combination produced something genuinely new in the history of the continuum traced so far: intelligence that began deliberately designing tools, and eventually began designing systems capable of performing elements of intelligence itself. Stone tools gave way to written language, written language to mathematics and formal logic, formal logic to mechanical calculation, and mechanical calculation, within a single human lifetime of the twentieth century, to the electronic computer. Each step remained squarely within human intelligence directing the process. The next step in the sequence — the one this paper is centrally concerned with — is different in kind, not merely degree.

8. The Continuum, Stage Four — Artificial Intelligence

Artificial intelligence, on the framework this paper proposes, is not an isolated invention arriving without precedent. It is intelligence designing another form of intelligence — the moment at which the continuum traced through Sections 5 through 7 begins, for the first time, to extend itself deliberately rather than through the blind processes of physical self-organisation or biological evolution. This is the paper's central claim, and it is worth restating in its most precise, most defensible form: artificial intelligence is humanity's first deliberate attempt to engineer a system capable of performing functions — pattern recognition, prediction, language processing, decision-making under uncertainty — that nature had previously only produced through billions of years of evolutionary search.

This framing carries a specific and, this paper argues, clarifying implication: AI is not best understood as replacing biological intelligence, in the way a calculator does not replace human arithmetic reasoning so much as extend what a person equipped with one can accomplish. Current AI systems remain narrow relative to the general adaptive flexibility of even simple biological organisms — a large language model can process and generate text with remarkable fluency while being unable to perform the basic environmental navigation an insect manages effortlessly. The continuum framework does not claim otherwise, and Section 12 addresses this asymmetry directly as one of the framework's most serious objections. What the framework does claim is narrower and, this paper argues, more defensible: that the underlying activity — engineered information processing, adaptation, and pattern extraction from data — belongs to the same broad category of phenomenon this paper has traced from physics through biology, engineered rather than evolved, and still extremely early in its own developmental trajectory relative to the multi-billion-year history of the biological intelligence that preceded it.

9. Why AI Is Becoming Infrastructure, Not a Tool

Electricity was once a novelty demonstrated at fairs. Within a few decades it disappeared into the background of ordinary life — nobody wakes up and consciously decides to "use electricity" before switching on a light. The internet followed a similar, if compressed, trajectory: a research curiosity in the 1970s, a visible novelty through the 1990s, and by the 2020s an invisible substrate underneath nearly every commercial and social activity, rarely remarked upon directly except when it fails.

This claim can be made more concrete than analogy alone, and doing so is worth the space it costs. In the United States, electricity reached roughly 8 percent of households by 1907 and did not cross 68 percent until 1929 — and took over three decades longer still to reach the majority of farm households after urban areas were already electrified, a genuinely slow, infrastructure-constrained diffusion. The personal computer reached early adopters in 1981 but did not reach a majority of American homes for another twenty years. Against that historical baseline, Anthropic's own Economic Index research finds that AI usage among US employees rose from 20 percent to 40 percent in the two years to 2025 — a pace the report notes took the internet, itself already considered a fast-diffusing technology, roughly five years to achieve. This is a genuinely testable, historically evaluable claim, not merely a rhetorical analogy, and the early data available as of this paper's writing is consistent with — though does not yet prove — the infrastructure trajectory this section proposes.

This paper's argument is that artificial intelligence is following that same basic trajectory toward invisible infrastructure, and that this is a natural consequence of its position on the continuum rather than a coincidence. Technologies that succeed in performing genuinely useful adaptive or information-processing functions tend to become infrastructure precisely because their usefulness does not depend on users consciously attending to the underlying mechanism — a claim consistent with, though not proof of, the framework advanced here. Future generations are unlikely to speak routinely of "using AI" in the way that framing already sounds slightly dated for electricity or the internet. They will simply perform everyday activities — medical diagnosis support, logistics routing, translation, drafting — supported by intelligent systems embedded so thoroughly throughout the relevant infrastructure that naming the AI component explicitly will start to feel as unusual as narrating which specific power plant is lighting a room.

10. Errors Are a Feature of Progress, Not an Exception to It

Every transformative technology in the industrial and post-industrial era has experienced significant, sometimes serious failures during its maturation, and every one of those technologies is still in routine, trusted use today. Medicines are recalled when post-market surveillance identifies risks that pre-approval trials did not catch. Food products are recalled when contamination is discovered. Aircraft receive mandatory safety directives when an operational pattern reveals a design or maintenance issue. Vehicles undergo recalls, sometimes for safety-critical defects, at a scale of millions of units per year across the global auto industry. Software is patched continuously, essentially as a permanent condition of its existence rather than a temporary phase.

Artificial intelligence and robotics are very unlikely to be an exception to this pattern, and this paper does not argue that they will be. The existence of AI failures — biased outputs, hallucinated facts, unsafe autonomous-system behaviour — should not, on the framework this paper proposes, be interpreted as evidence against the underlying legitimacy of the technology. It should be read the way failures in every other listed technology are read once enough time has passed: as the ordinary, expected texture of an engineering discipline still maturing its regulatory apparatus, its testing methodology, and its failure-response infrastructure. This is a claim about how to interpret failure, not a claim that any given AI failure is acceptable, minor, or exempt from scrutiny — it is precisely the opposite: mature engineering disciplines are defined by how rigorously they respond to failure, not by the absence of failure.

11. Hybrid Intelligence — The Proposed Next Stage

This paper's framework points toward a specific claim about what comes after the current stage of the continuum, and it is important to be precise that this claim is not being coined here for the first time. Hybrid Intelligence is an established term in the academic human-computer interaction and information systems literature, formally defined by Dellermann, Calma, Lipusch, and colleagues in a 2019 paper as "the ability to achieve complex goals by combining human and artificial intelligence, thereby reaching superior results to those each of them could have accomplished separately, and continuously improve by learning from each other." A related formulation by Akata and colleagues frames it as AI amplifying human intellectual capability rather than replacing it, oriented toward decisions and actions neither humans nor machines could reach alone. AlphaGo is frequently cited in this literature as an early, concrete example: an AI system that both learned from an enormous corpus of human game data and subsequently taught human players genuinely novel strategic ideas, a two-directional flow of capability rather than a one-directional replacement.

This paper adopts that existing definition rather than proposing a competing one, and situates it as the next natural stage of the continuum traced through Sections 5 through 8. If AI is best understood as engineered information processing built on top of, rather than separate from, the biological intelligence that came before it, then a stage in which human and artificial intelligence operate as a genuinely collaborative system — rather than as separate, competing categories — is the more natural extrapolation of the continuum's trajectory than a scenario in which one form of intelligence fully supersedes or replaces the other. This has direct, practical implications across medicine, education, scientific discovery, manufacturing, transportation, and public services, where the most successful current deployments of AI overwhelmingly take the augmentation form the Hybrid Intelligence literature describes, rather than the full-autonomy form popular discourse often assumes is the technology's inevitable end state.

It is worth stating this derivation explicitly rather than leaving it implicit, since the inference does real work in this paper's argument. Section 1 established that each stage of the continuum layers on top of, rather than replaces, the stage before it: chemistry did not stop once biology emerged, and biological adaptation did not stop once human reasoning appeared. If that layering pattern is a genuine structural feature of the continuum rather than a coincidence of the four stages examined so far, the most consistent extrapolation is that engineered intelligence will similarly layer onto, rather than fully replace, the human intelligence that engineered it — which is a structural description of what Hybrid Intelligence actually is. This derivation is a conceptual inference from a pattern observed across four prior stages, not a proof that the pattern must continue into a fifth, and this paper treats it accordingly: as the framework's most defensible prediction, not its most certain one.

12. Objections and Counterarguments

A framework this broad invites serious objections, and a paper that does not engage them directly has not actually defended its thesis. Four are addressed here specifically because they are the strongest available, not the easiest.

12.1 The Specified-Information Objection

As flagged in Section 4.2, a genuine scientific and philosophical debate exists over whether self-organisation alone can account for the specified, code-like information found in biological systems such as DNA — information that behaves less like the spontaneous pattern-formation Prigogine and Kauffman documented and more like the specified, functional information found in written language or software, a distinction pressed seriously by information theorist Hubert Yockey and others. This paper's response is not to resolve that debate, which remains genuinely open within the relevant scientific literature, but to note that the continuum framework does not actually require it to be resolved in any particular direction. Whether biological information-storage capacity arose purely through self-organising physical processes or required additional explanatory mechanisms, the observable fact this paper relies on — that biological systems do store, transmit, and act on information, by whatever process that capacity arose — remains true either way.

12.2 The Category Error Objection

A second, sharper objection holds that grouping thermodynamic self-organisation, immune-system pattern recognition, human abstract reasoning, and large language model text generation under a single "intelligence" umbrella is not illumination but definitional dilution — that the word is being stretched until it means everything and therefore nothing. This is the single strongest objection to this paper's framework, and it is only partially answerable. This paper's response is that the continuum is explicitly graded, not flattened: it does not claim a thermostat and a human being possess equivalent intelligence, only that both sit somewhere on a single continuum of information-processing sophistication, with enormous, explicitly acknowledged distance between the two points. Readers who find even this graded version of the claim too permissive a use of the word "intelligence" are registering a genuine and reasonable disagreement with this paper's central organising choice, not a misunderstanding of it.

One further point deserves stating rather than leaving implicit: the choice to treat intelligence as graded and continuous, rather than as a set of discrete categories with sharp boundaries, is itself a philosophical commitment this paper is making, not a neutral description of how the world actually is. A framework built instead on discrete categories — physical systems, biological systems, human cognition, and engineered systems treated as genuinely separate kinds rather than points on one continuum — is equally coherent and defensible, and would simply be a different paper with a different, likely less unifying, set of conclusions. This paper's choice of the graded continuum is defended on explanatory grounds throughout Sections 4 through 8, not claimed as the only reasonable way to organise the same underlying facts.

12.3 The Consciousness Objection

A third objection notes that this paper spends considerable effort distinguishing information processing from subjective experience in Section 3.3, and then asks why that distinction should be treated as decisive rather than as begging the actual question anyone asking about "intelligence" cares about — namely, whether anything is experiencing this processing from the inside. This is a fair challenge. This paper's honest answer is that the hard problem of consciousness remains unresolved in philosophy of mind generally, well beyond any question specific to AI, and that this paper's continuum framework is deliberately scoped to avoid depending on its resolution. That scoping is a limitation of this paper, not a solution to the underlying philosophical problem, and readers primarily interested in machine consciousness specifically should treat this paper as declining to address their central question rather than as answering it.

12.4 The Legitimation Objection

A fourth objection is less philosophical and more political: that framing AI as the latest stage of a natural continuum risks functioning as rhetorical cover for under-regulating a technology with genuine, present-day harms — algorithmic bias, labour displacement, concentration of power among a small number of AI developers — by making its arrival sound inevitable and its trajectory sound benign. This is a legitimate concern about how this kind of framework could be misused, independent of whether the framework itself is correct. This paper's position is that "this technology has deep precedent" and "this technology requires rigorous regulation" are not in tension — electricity and the internet both proved to have deep historical continuity with prior infrastructure and still required, and continue to require, substantial regulatory frameworks to govern their harms. Nothing in this paper's continuum argument implies AI should be regulated more lightly than its actual risks warrant, and Section 13 states that directly rather than leaving it as an inference.

13. Implications for Policy, Business, and Society

If the continuum framework holds even loosely, several practical implications follow that differ from the implications of the discontinuity narrative this paper opened by critiquing. One point from Section 12.4 is worth restating here, prominently, before the specific implications below, because it is the point most likely to be misread if left only in the objections section: deep historical continuity is not an argument for lighter-touch regulation. A technology having deep precedent says nothing about how much oversight its present-day risks warrant; those are separate questions, and this paper answers only the first.

For policy: regulation should be modelled on how other infrastructure-stage technologies were governed as they matured — continuous, adaptive, and responsive to demonstrated harms, the pattern applied to electrical safety codes and telecommunications standards — rather than either a one-time settlement or a permissive stance justified by the technology's supposed novelty.

For business: the Hybrid Intelligence literature's finding that augmentation deployments consistently outperform full-autonomy deployments in current practice suggests organisations should generally prioritise AI implementations that combine human judgement with AI capability, rather than architectures aimed at fully removing human involvement from a given process.

For public understanding: a framework that treats AI as continuous with a long history of adaptive systems, rather than as an unprecedented rupture, may help calibrate public expectations away from the two extremes — utopian inevitability and existential rupture — that currently dominate popular discourse, toward a more measured expectation of a genuinely significant but non-magical technology maturing along a recognisable historical pattern.

14. Limitations

This paper's limitations are substantial and are stated here directly rather than minimised.

• The central framework is philosophical and interpretive, not empirical, and this deserves the plainest possible statement: it is not predictive, and it does not generate falsifiable hypotheses that could, in principle, prove the continuum framework wrong. It organises established science into a narrative; it does not itself constitute a testable scientific theory, and readers should not treat it as one. This is a genuine limitation, not a strength dressed up as humility — a framework immune to disconfirmation is harder to take seriously as an explanatory tool, not easier, and this paper does not claim otherwise.

• The breadth of the continuum — from thermodynamics to large language models — necessarily sacrifices depth and precision at each individual stage. Specialists in any one of the fields touched (complexity science, immunology, cognitive science, machine learning) will reasonably find their own area treated at a level of simplification a dedicated paper in that field would not permit.

• Section 12.1's specified-information debate is genuinely unresolved in the underlying scientific and philosophical literature, and this paper's decision to treat the continuum framework as independent of that resolution is a scoping choice, not a demonstration that the choice is costless.

• Section 12.2's category-error objection is, in this paper's own assessment, not fully answerable, only partially answerable. Readers who reject the graded-continuum use of "intelligence" are not wrong to do so; this represents a genuine, unresolved disagreement about definitional scope rather than a misreading correctable by clarification.

• This paper does not address machine consciousness or the hard problem of consciousness, by deliberate scope decision stated in Section 3.3 and revisited in Section 12.3, and should not be read as taking an implicit position on either question.

• The paper's claim that AI is becoming infrastructure in the manner of electricity and the internet (Section 9) is an analogy, not a proof, and analogies of this kind can fail for reasons specific to the technology being compared — a risk this paper acknowledges without being able to fully resolve in advance of the outcome actually unfolding.

15. Conclusion

Artificial intelligence is not the beginning of intelligence. Nor is it necessarily its final form. This paper has argued that it is one recent chapter in a much longer story: the gradual, four-billion-year emergence of increasingly capable systems that sense, adapt, store information, and — for the first time in this stage of the continuum — deliberately engineer further intelligence rather than simply evolving it. Humanity now stands at the point in that continuum where intelligence has begun designing intelligence itself, a genuinely significant moment, though this paper has argued throughout that its significance is better understood as continuity with deep history than as rupture from it.

This framework will not resolve the genuine scientific debates named in Section 12, nor should it attempt to. What it offers instead is a way of situating artificial intelligence within a much longer, better-documented history of adaptive systems — one this paper argues gives AI's own trajectory, including its failures, its regulatory needs, and its likely destination in some form of Hybrid Intelligence, a more stable and more defensible interpretive foundation than the discontinuity narratives this paper opened by questioning.

One closing point is worth restating rather than trusting readers to recall from Section 3.3: nothing in this paper claims the universe itself is intelligent or conscious, and nothing in it claims that present-day artificial intelligence systems have subjective experience. The continuum described here is a continuum of information processing and adaptive capability, observed and argued from evidence and precedent. What it feels like, if anything, to be any point on that continuum other than the human one remains exactly as open a question at the end of this paper as it was at the beginning.

16. References

Wiener, N. (1948). Cybernetics: Or Control and Communication in the Animal and the Machine. MIT Press.

Shannon, C. E. (1948). A mathematical theory of communication. Bell System Technical Journal, 27(3), 379-423.

McCulloch, W. S., & Pitts, W. (1943). A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics, 5(4), 115-133.

Ashby, W. R. (1957). An Introduction to Cybernetics. Chapman & Hall.

Nicolis, G., & Prigogine, I. (1977). Self-Organization in Non-Equilibrium Systems: From Dissipative Structures to Order Through Fluctuations. Wiley.

Kondepudi, D., & Prigogine, I. (2014). Modern Thermodynamics: From Heat Engines to Dissipative Structures (2nd ed.). John Wiley & Sons.

Kauffman, S. A. (1993). The Origins of Order: Self-Organization and Selection in Evolution. Oxford University Press.

Kauffman, S. A. (1995). At Home in the Universe: The Search for Laws of Self-Organization and Complexity. Oxford University Press.

Haken, H. (1988). Information and Self-Organization: A Macroscopic Approach to Complex Systems. Springer-Verlag.

Bak, P., Tang, C., & Wiesenfeld, K. (1987). Self-organized criticality: An explanation of the 1/f noise. Physical Review Letters, 59(4), 381-384.

Yockey, H. P. (2005). Information Theory, Evolution, and the Origin of Life. Cambridge University Press.

Dellermann, D., Calma, A., Lipusch, N., Weber, T., Weigel, S., & Ebel, P. (2019). The Future of Human-AI Collaboration: A Taxonomy of Design Knowledge for Hybrid Intelligence Systems. Proceedings of the 52nd Hawaii International Conference on System Sciences.

Akata, Z., Balliet, D., de Rijke, M., et al. (2020). A research agenda for hybrid intelligence: Augmenting human intellect with collaborative, adaptive, responsible, and explainable artificial intelligence. Computer, 53(8), 18-28.

Silver, D., Huang, A., Maddison, C. J., et al. (2016). Mastering the game of Go with deep neural networks and tree search. Nature, 529(7587), 484-489.

Anthropic. (2026). Anthropic Economic Index: Uneven geographic and enterprise AI adoption. (Includes comparative adoption-speed data for AI, internet, electricity, and personal computers.)

Darwin, C. (1880). The Power of Movement in Plants. John Murray. (Foundational early observation of phototropism.)

Mueller, W. C., & Greenberg, J. T. (2022). Plant immunity: The origami of receptor activation. Current Biology, 32(9), R418-R420.

Wiltschko, W., & Wiltschko, R. (2005). Magnetic orientation and magnetoreception in birds and other animals. Journal of Comparative Physiology A, 191(8), 675-693.

Burbidge, E. M., Burbidge, G. R., Fowler, W. A., & Hoyle, F. (1957). Synthesis of the elements in stars. Reviews of Modern Physics, 29(4), 547-650.

About the Author

Pawan Bhatia is the founder of NextGen Economics, an independent research platform publishing long-horizon analysis on capital, technology, and the forces shaping the next decade of wealth creation.

His work is guided by a singular vision: a world in which the quality and distribution of economic thinking is radically improved — a 1Q world, in which better questions lead to better outcomes.

This paper was built with the assistance of AI tools for research synthesis, drafting, and formatting. All arguments, framings, and editorial decisions are the author's own.

Contact: pawan.bhatia@nextgeneconomics.com · www.nextgeneconomics.com

Publisher's Note

NextGen Economics is an independent research platform dedicated to producing evidence-based, accessible, and actionable research on the societal implications of technological change. Our mission is to bridge the gap between emerging technologies and economic governance. We are committed to intellectual independence. No corporate funding, advisory relationships, or consulting engagements influenced the analysis or conclusions presented in this document.

Disclaimer

This white paper presents a philosophical and interpretive framework, not an empirical scientific finding or a statement of settled fact. The views expressed herein are those of the author and do not necessarily reflect the views of any affiliated organisations. This document does not constitute scientific, medical, legal, or regulatory advice.

AI Tools Disclaimer: This white paper was developed with the assistance of AI-powered research and writing tools, including large language models for literature synthesis, drafting, and formatting. All substantive arguments, philosophical framings, and final editorial decisions remain the sole responsibility of the author. The use of AI tools does not imply endorsement of the content by any AI system or its developers.

Reproduction and Distribution: This white paper is licensed under Creative Commons Attribution-NonCommercial 4.0 International License. You are free to share, adapt, and distribute this work for non-commercial purposes, provided you attribute the author appropriately.