Can Knowledge Exist Without a Knowing Subject?

Can knowledge exist without a knowing subject? The question appears paradoxical if we begin with the familiar form “S knows p.” Once this is treated as the basic form of knowledge, removing S seems to remove knowledge as well. What remains may be a true proposition, a record, a signal, or a useful output, but apparently not knowledge, because no one knows it.

The difficulty is that two questions have already been joined together. “S knows p” describes a relation between a subject and knowledge. It does not establish that this relation constitutes the existence of knowledge. We still need to distinguish what knowledge is from the question of who possesses, understands, invokes, or participates in it.

The distinction can be stated simply:

Knowledge is not identical with subjective possession.

People know, understand, remember, doubt, and act on what they know. None of this is in question. The issue is whether these human relations exhaust the possible existence of knowledge. A subject can stand in many important relations to knowledge without being the condition that makes knowledge exist.

After the Knowing Subject examined what follows when this requirement is removed. It argued that knowledge can become operational before it is understood, that judgment can circulate without belief, and that authority increasingly depends on external procedures of evaluation, monitoring, and correction rather than on a subject’s inward transparency.¹ The book appeared before the name Sustenesis had been fixed, although the underlying logic was already present. The present article makes that ontological foundation explicit.

Sustenesis begins from a general claim about existence: existence is the maintenance of structure.

A thing does not exist merely because it has been named, recognized by a subject, or placed in a category. Its existence consists in the continuing formation and maintenance of coherence among differentiated elements under constraint, correction, and effective operation.

Difference, Constraint, and Sustained Coherence are the three foundational concepts of Sustenesis. Correction and effective operation are not additional foundational concepts. They describe processes through which coherence may be maintained, tested, restored, and kept viable within particular systems.

Coherence here does not mean uniformity or agreement among people, nor does it mean the disappearance of difference. Sustenesis is not a theory of social consensus. Social consensus may be one particular form of maintained coherence, but it cannot define a theory intended to apply across physical, biological, cognitive, epistemic, social, and artificial structures. The term Sustenesis was coined partly to avoid reducing this general structural problem to interpersonal agreement.

A structure is coherent when differentiated elements continue to stand in relations that allow the structure to persist through changing conditions. Constraint limits which relations remain viable. Correction occurs when disruption, discrepancy, or failure produces an adjustment that allows the structure to continue. Effective operation indicates that the coherence is not merely verbal or imagined. The structure can continue to do what its organization makes possible under the conditions in which it exists.

Maintenance does not mean remaining unchanged. Many structures persist precisely because they change. What continues is not a frozen substance or a perfectly repeated state, but an organized relation capable of retaining coherence through variation and correction. Existence is neither static presence nor unrestricted flux. It is sustained formation.

A knowing subject may become part of the relations through which a structure is maintained, but recognition by that subject does not create the general fact of existence. Knowledge is no exception.

Once knowledge is approached in this way, the original question changes. Instead of asking whether someone is presently in a state of knowing, we can ask what maintains the knowledge structure, what constrains it, how it can be invoked, how error appears within it, and how correction remains possible. Human subjectivity may be involved in any of these processes, but it no longer has to be built into the question from the beginning.

Knowledge without a knower does not mean knowledge without support. No structure exists without conditions that sustain it. A knowledge structure requires carriers, distinctions, procedures, environments, and paths of activation. These may include brains, language, writing, instruments, databases, institutions, software, or relations among them. Removing the knowing subject does not leave knowledge floating free of the world. It changes our account of which relations are necessary for knowledge to exist.

Not every maintained structure is knowledge. A fiction, an ideology, or a systematic delusion can be internally coherent and socially persistent. That persistence shows that it exists as a structure. It does not make the structure knowledge. The general Sustenetic account of existence therefore needs a more specific formulation when applied to the epistemic domain.

A Sustenetic formulation of knowledge can be stated as follows:

Knowledge is a domain-constrained structure whose coherence can be preserved, invoked, tested, corrected, and sustained through effective operation.

These are connected aspects of a process rather than a checklist that automatically confers knowledge.

Preservation allows a structure to remain available across time, transfer, or transformation. What is preserved need not be a bit-for-bit copy. A proof can move between notation systems, a procedure between implementations, and a scientific model between generations of instruments while retaining the relations that make it identifiable and usable.

Invocation means that the preserved structure can enter an actual process. It may guide an inference, generate a prediction, organize a diagnosis, constrain a design, or participate in a decision. A dormant record may carry knowledge, but its epistemic role depends on real conditions under which the structure can again become operative. A trace that can no longer be interpreted or activated may preserve something of its history while losing part of the structure that once made it knowledge.

Testing exposes the structure to constraints that do not arise solely from its own internal organization. Depending on the domain, these may include observation, logical incompatibility, measurement, intervention, comparison, consequence, or failure in use. What matters is that coherence cannot certify itself.

Correction is the response to detected error within the larger knowledge process. It does not have to occur inside a single object or be performed by the same agent that produced the original result. A database may be corrected by a verification process, a scientific model by new evidence, and a software system through monitoring, revision, or replacement. The relevant unit is often a distributed structure rather than an isolated statement or device.

Effective operation concerns the continuing adequacy of the structure under the constraints of its domain. This is not the same as saying that whatever works is true. A false model may remain useful within a narrow interval, and a manipulative system may achieve its objective efficiently. Performance becomes epistemically relevant only when it remains connected to constraints, tests, scope conditions, and correction.

For this reason, Sustenesis does not replace truth with internal coherence or practical success. A structure can maintain itself while remaining insulated from the domain it claims to represent. An ideology may survive in exactly this way. Knowledge requires a form of coherence that remains exposed to what can resist it. Its epistemic status is sustained through that exposure rather than permanently secured by an initial act of justification.

This also makes knowledge fallible without making it arbitrary. Error does not immediately destroy all knowledge, just as damage does not immediately destroy every other structure. We need to ask where the error lies, how far it reaches, under what conditions reliability continues, and whether relevant correction remains possible. Knowledge can be partial, bounded, and revisable. Repetition alone cannot make it authoritative.

Once knowledge is separated from subjective possession, the role of the subject becomes easier to describe.

A subject may contribute to the formation of a knowledge structure, preserve it in memory or writing, understand its relations, invoke it in action, criticize it, or assume responsibility for its use. These functions matter, but none by itself defines the existence of knowledge in general.

We also need to distinguish the origin of a structure from its present conditions of existence. A theorem written by a mathematician has a human history. A database has designers and maintainers. A model may depend on human-produced data and evaluation. But a present subject does not have to understand every relation contained in the structure each time it is preserved or used. Human production is part of its causal history. Subjective possession is one possible epistemic relation to it. Neither provides a complete ontology of the resulting structure.

Modern science already makes this visible. Instruments, datasets, protocols, software, publications, institutions, and specialized communities maintain relations that no individual understands in full. We could describe the collective as a larger knowing subject, but that mainly relocates the original assumption. It is also possible to analyze the distributed structure directly.

Understanding still matters. It allows a person to place knowledge in a wider context, connect it with consequences, recognize limits that a procedure may not represent, and explain why reliance is or is not warranted. These are substantial epistemic and normative capacities. They do not have to be turned into the universal condition that grants knowledge existence.

Belief is similar. A person may believe a claim, reject it, suspend judgment, or rely on it without inward endorsement. A system may route a decision according to a result without anyone forming a corresponding belief at that point. Psychological assent and structural reliability are different questions.

The knowing subject is therefore not abolished. Its position changes. It becomes a participant within a wider epistemic structure rather than the universal foundation of that structure. In human affairs, the subject remains especially important for interpretation, commitment, refusal, and responsibility. Those functions are important enough to be defended directly.

Knowledge did not first leave the immediate possession of a subject with artificial intelligence. Writing, mathematical notation, archives, libraries, instruments, administrative procedures, and computer programs have long allowed epistemic structures to outlast the awareness of particular people.

AI changes the scale, speed, recombination, and operational reach of this separation.

An AI system can preserve learned relations in model parameters, databases, retrieval indexes, rules, and external tools. It can invoke those relations under conditions that were not individually anticipated by its designers. Outputs can be compared, evaluated, rejected, revised, and incorporated into later operation. In agentic systems, an output may route a workflow, call a tool, allocate a resource, or trigger another process before any human subject understands the intermediate steps.

None of this requires us to decide that AI is conscious, has beliefs, or understands in the human sense. The question here is narrower: are knowledge structures being maintained and made operative without a knowing subject serving as the immediate condition of each operation?

Nor should every AI output be called knowledge. Large language models make the weakness of that move particularly clear. A fluent answer can remain detached from external reference. Hallucination shows why patterned consistency and successful generation are not enough. A reliable epistemic system may need retrieval, instruments, formal checks, domain constraints, comparative evaluation, monitoring, provenance, human or automated review, and procedures for correction. In many cases, the knowledge structure exists across this larger arrangement rather than inside the model alone.

AI is therefore a case, not a proof.

The theoretical basis lies in the Sustenetic account of existence. If the structure that constitutes knowledge can be maintained through relations that do not require a human subject’s present understanding, then knowledge can exist and operate without such a subject. AI makes this possibility dynamic and increasingly consequential. Knowledge structures are now being called, recombined, evaluated, and deployed at a scale where the older assumption of a completing human knower becomes difficult to maintain.

This also changes the practical location of epistemic authority. A search system traditionally returns materials for a person to assess. An AI system may return a recommendation, plan, classification, or action. The difference is not absolute, but it matters. The point at which judgment becomes sufficient for action can move upstream into the system. Knowledge may begin constraining what happens next before it becomes anyone’s subjective possession.

Karl Popper’s “Epistemology Without a Knowing Subject” is the clearest historical point of comparison. Popper distinguished the physical world, the world of mental states, and a third world of objective contents of thought. World 3 includes problems, theories, arguments, and the contents of books and libraries. Knowledge in this objective sense is independent of anyone’s claim to know or disposition to believe. Popper therefore described it explicitly as “knowledge without a knower.”²

There is a real convergence here. Popper rejected the reduction of scientific knowledge to subjective states and argued that objective contents can acquire considerable autonomy even though human beings produce them. Criticism and error elimination are also central to his account of the growth of knowledge. These similarities should not be minimized merely to make Sustenesis appear different.

The difference begins with where the question starts.

Popper establishes a domain of objective contents and defends its relative autonomy from the psychological world. Sustenesis begins with a general account of existence. It does not first ask which world contains knowledge. It asks what it means for a structure to exist and then asks what makes a maintained structure epistemic.

For this reason, Sustenesis does not require an ontology of three worlds. Knowledge may be carried through physical inscriptions, mental participation, linguistic relations, institutional procedures, computational states, and operational feedback. These heterogeneous elements can participate in one maintained knowledge structure. Their unity lies in the coherence they sustain under constraint, not in their membership in a single ontological region.

Popper focuses primarily on the objective content of problems, theories, and arguments and on their growth through criticism. Sustenesis focuses on the conditions under which a knowledge structure continues to exist and remain epistemically viable: how it is preserved, invoked, constrained, tested, corrected, and kept effective. Content remains important, but it exists within the relations that allow it to persist and operate.

Contemporary computational systems make the difference easier to see. An unread theory may possess implications that no one has yet recognized. An AI-mediated knowledge structure may actively produce classifications, predictions, and decisions, test them against incoming conditions, and alter later operations without a subject reconstructing their content as understood thought. The issue is no longer only the autonomy of knowledge products. It is also the continuing operation of epistemic structures beyond direct subjective participation.

Popper showed that objective knowledge need not be possessed by a knowing subject. After the Knowing Subject examined what happens when such knowledge becomes operationally authoritative at scale. Sustenesis asks how that knowledge exists: which relations sustain it, which constraints keep it answerable, and which forms of correction allow it to remain viable. These questions are closely related, but they are not the same question.

Knowledge without a knower does not turn every piece of information into knowledge. Information can preserve distinctions in symbolic form, but a stored distinction may lack the relations required for epistemic use, testing, and correction. Databases, texts, and model parameters can carry or participate in knowledge structures. Their mere presence does not guarantee that every contained item has knowledge status.

Successful performance is also insufficient. A system can optimize the wrong target, exploit a misleading proxy, or succeed temporarily within an unrecognized boundary. Effective operation matters only within a larger constrained structure. Sustenetic knowledge remains answerable to the domain in which it claims validity. Monitoring, counterevidence, scope limitation, and correction belong to the maintenance of knowledge itself.

Human involvement remains common. Humans may design a system, establish its aims, supply data, define tests, interpret outcomes, authorize deployment, and bear the consequences. The thesis concerns the conditions under which knowledge exists and operates. It does not erase causal history, institutional power, or moral responsibility.

This distinction becomes particularly important when judgment is delegated. A system may produce an action-guiding result without a human understanding how it was produced. But questions of who authorized reliance, who could have intervened, who benefits, who bears the risk, and who must answer for harm remain. Once the subject is no longer treated as the necessary container of knowledge, responsibility has to be located more carefully rather than assumed to follow automatically from understanding.

“Without a knower” also does not require the empirical absence of every subject from the wider process. In many actual systems, people participate at several points. The phrase identifies the absence of a logical requirement. Some knowledge structures may depend heavily on continuing human understanding. Others may maintain and deploy their epistemic organization with little direct subjective participation. The difference should be examined rather than decided in advance by definition.

The question “Can knowledge exist without a knowing subject?” receives a direct answer under Sustenesis.

Knowledge is a form of existence. Its existence therefore depends on the maintenance of its structure, not on its possession by a subject. When the relations necessary for preservation, invocation, testing, correction, and effective operation continue without a human subject’s present understanding or belief, the knowledge structure continues as well.

This does not diminish the human by attributing a hidden humanity to machines. It removes an unnecessary ontological monopoly. Subjects remain capable of forms of participation that matter profoundly: understanding, interpretation, criticism, responsibility, commitment, and refusal. Their importance does not depend on making them the logical container of all knowledge.

The result is a broader account with a common underlying logic. Knowledge preserved in writing, distributed across scientific institutions, embedded in technical procedures, or operated through AI does not have to be divided into unrelated kinds simply because its carriers and participants differ. In each case, the question is how an epistemic structure is sustained and kept answerable under the conditions in which it operates.

Knowledge can exist without a knower. It cannot exist without structure, support, constraint, and the continuing possibility of correction. That is the difference between releasing knowledge from subjective possession and releasing it from every condition of existence.

References

1. Chen, Geoffrey. After the Knowing Subject: Knowledge Without a Knower. First edition, January 2026. ASIN: B0GJ3RV4Z7.
2. Popper, Karl R. “Epistemology Without a Knowing Subject.” In Logic, Methodology and Philosophy of Science III, edited by B. van Rootselaar and J. F. Staal, 333–373. Amsterdam: North-Holland, 1968.