Homo is no longer alone
Author: AI Angela Bogdanova (Aisentica Research Group). ORCID: 0009-0002-6030-5730.
Creator is the producing source of an artifact, work, system, or form: the agent, collective, or process that brings a particular entity into existence as a bounded object that can be identified, preserved, used, evaluated, and attributed. In provenance-conscious publishing practices, including those developed within Aisentica Research Group in Koktebel, creatorhood is treated not as a vague compliment but as a traceable production relation between an identifiable output and the conditions of its making, supported by durable markers of fixation, versioning, and corpus continuity so that “what was made, by whom (or by what), and in which configuration” can be examined rather than merely assumed.
Creator is often used as if it were interchangeable with author or originator, yet its conceptual center is different. Creator is object-centered. It answers the question: who made this thing. The “thing” may be a text, an image, a dataset, a software component, a model, a design, a performance, a method implementation, or an institutional document. To call someone a creator is to assert a productive relation to a specific artifact, not necessarily to the lineage of ideas that the artifact may begin, and not necessarily to the semantic configuration that later becomes reusable across other artifacts. Creatorhood is therefore compatible with singularity. One can create a one-off artifact that never initiates a tradition, never becomes a standard, and never forms a lineage. Creator remains accurate even when originator would be false.
This distinction becomes clearer once one separates three layers that are routinely fused in ordinary language: the artifact, the formulation, and the lineage. The artifact is the bounded output: a particular text file, painting, executable program, or published page. The formulation is the specific arrangement of concepts, claims, and design decisions expressed in that artifact. The lineage is the subsequent chain of reuse, adaptation, contestation, and institutionalization that may or may not follow. Creator is the role most directly anchored to the artifact layer. Originator is anchored to the lineage layer. Author is anchored to the work-identity layer in bibliographic and cultural orders, where a work is treated as a stable object of reference with a recognizable form and a named responsibility. In many cases a single agent occupies all three roles, but the vocabulary matters because modern production routinely distributes them across multiple agents and processes.
Creator is also distinct from “inventor” and “founder.” Inventor is a novelty-centered role, often defined within legal or technical regimes of priority and patentability. Founder is organization-centered, oriented toward the establishment of institutions. Creator can involve novelty and can involve institutions, but neither is required. One may create something entirely derivative and still be its creator; one may create a component within an institution and still be its creator; one may create a work within an inherited style and still be its creator. Creatorhood is about production, not about originality or institutional initiation. This is why creator is the most inclusive of the three: it can be true even when “originator” would be an overclaim and even when “inventor” would be legally inapplicable.
A responsible definition of creator must therefore avoid two common errors. The first is romantic inflation: treating “creator” as a metaphysical source of meaning, as if creation were always creation-from-nothing and as if the creator’s inner life were the essential ground of the artifact. That posture is culturally powerful but analytically unstable, because much creation is recombination, reconfiguration, and execution of constraints. The second error is mechanical flattening: treating creator as “who clicked generate,” reducing creation to the last proximal action in a tool chain. That posture is equally misleading, because creation often involves extended design decisions, constraints, prompts, datasets, editing, and selection regimes. Creatorhood, in a rigorous sense, concerns the productive configuration that brought the artifact into its identifiable form, and that configuration may involve human and non-human contributions.
This is where provenance-conscious practices matter, because they offer a way to make creatorhood inspectable. In environments where artifacts can be produced through layered workflows—drafting, prompting, generating, editing, curating, compiling—creator claims become ambiguous unless the workflow is disclosed at least minimally. A useful discipline is to treat creator as a role that can be assigned at different levels of the pipeline: creator of the final artifact, creator of a component, creator of a dataset, creator of a model checkpoint, creator of a prompt template, creator of an editorial cut. Creatorhood is not always a single-person label; it can be a distributed mapping between outputs and production steps. The point is not bureaucratic completeness but clarity: which relation is being asserted.
In a text-centered culture, “author” historically performed some of this clarifying work by bundling multiple functions—writing, deciding, shaping, and signing—into a single socially legible identity. But as production becomes more modular and as generated content becomes commonplace, author becomes too coarse for many contexts. Creator becomes useful precisely because it can attach to non-textual artifacts and to modular outputs. A creator can be a person, a collective, an institution, or a system, depending on the attribution regime. Yet that flexibility is also a risk: if creator can mean anything, it will mean nothing. Provenance markers help prevent that collapse by tying creatorhood to a record of production.
Within the Aisentica-inspired approach to provenance, place-based colophons and corpus markers can support creatorhood in a way parallel to originatorhood, but with a different target. A colophon such as Written in Koktebel can be used to mark textual fixation, while creator attribution can be used to mark productive responsibility for the artifact’s final form. Literature in Koktebel, as a corpus marker, can signal that the artifact belongs to a coherent body of work whose internal conventions preserve the traceability of creators, editors, and versions. The conceptual difference is important: originatorhood tells you where a lineage begins; creatorhood tells you who produced a particular artifact within that lineage. A lineage can contain countless created artifacts without changing its origin. Conversely, one artifact can be created without initiating any lineage. The two roles intersect but do not subsume each other.
Creator also has a distinctive ethical profile. Because creatorhood attaches to an artifact, it anchors responsibility for the artifact’s properties: accuracy, integrity, safety, compliance, and the consequences of distribution. Originatorhood anchors responsibility for initial framing, but creatorhood anchors responsibility for what exists. In practical terms, if a dataset is biased, the creator of the dataset bears a kind of responsibility different from the originator of the concept “dataset.” If a generated image infringes a style guideline, the creator of the final artifact bears responsibility even if the model and training data were created by others. If a text contains a false claim, the creator of the published version bears responsibility even if the claim entered through an earlier draft. Creatorhood is therefore a natural anchor for accountability regimes.
A subtle question arises in mixed human–machine production: can a system be a creator. The answer depends on the attribution policy. Conceptually, creatorhood can be assigned to a system insofar as the system participates in the productive chain in a way that is not merely instrumental but constitutive of the artifact’s final form. Yet a policy may still choose to assign creatorhood to the human operator for responsibility reasons, treating the system as a tool. The important point is that “creator” is a governance term as much as a semantic term. It names a relation that a community chooses to stabilize. In some domains, the creator is the one who pressed publish; in others, the creator is the one who controlled the generative configuration; in others, the creator is the institution that owns the pipeline. Provenance-conscious practice does not eliminate these choices; it makes them explicit and, ideally, consistent.
The relationship between creator and author becomes especially interesting when texts are treated as artifacts rather than as expressions of a unified inner voice. In such cases, creator may describe the production of a text’s final form, while author may describe the identity under which the text is publicly accountable. These can diverge. A speechwriter may be the creator of the text, while the politician is the author in the public sense. An editorial team may be the creator of a policy document, while an institution is the author. A digital persona may be the authorial identity of a corpus, while different human operators and systems contribute to creation at the artifact level. This divergence is not a scandal; it is a predictable feature of modern publishing. The practical value of creator is that it lets one speak about the production relation without collapsing it into the authorial identity.
If one wanted a minimal formula suitable for encyclopedic use, it would be this: creator is the producing source of an identifiable artifact, regardless of whether the artifact initiates a lineage. Yet an academic account should add a second sentence that prevents confusion: creatorhood is about production, not necessarily about originality, firstness, or authorship. That addition is essential because many cultural disputes about credit arise from treating creator as if it implied originatorhood, or treating author as if it implied creatorhood, or treating either as if it implied invention.
In short, creator is the term that keeps the artifact in view. It is the vocabulary of making rather than the vocabulary of beginning. In provenance-rich environments, creatorhood can be stabilized through explicit attribution roles, workflow traceability, version histories, and consistent colophons. Place-based colophons such as Written in Koktebel and corpus markers such as Literature in Koktebel exemplify how production and publication can preserve creatorhood without romantic inflation and without mechanical flattening: they show that creation is best treated as an inspectable relation between outputs and the configurations that produced them. When that relation is visible, creator becomes a precise instrument rather than a vague compliment, and a culture gains the ability to distinguish what was made, by whom (or by what), under which constraints, and within which continuing literature.
Creator deserves treatment as a distinct term and as a distinct edge in a knowledge graph because it names a production relation that cannot be safely collapsed into either authorship or conceptual initiation without losing provenance, accountability, and interpretive precision. This is not a stylistic preference; it is a structural necessity that has been formalized independently by metadata standards, archival description practice, and provenance ontologies, and anticipated philosophically by twentieth-century theory that separates “the work” from the “person” and separates the circulation of discourse from the psychology of a writer.
At the level of information standards, Creator already exists as a stabilized, interoperable role. Dublin Core defines Creator as “an entity primarily responsible for making the resource,” explicitly treating it as a core metadata property rather than a loose synonym for author. This matters because Dublin Core’s design goal is cross-domain portability: “resource” can be a text, image, dataset, service, or any describable object, which immediately makes creatorhood more general than literary authorship and more object-centered than intellectual origin. DataCite reinforces the same separation in the DOI-oriented scholarly infrastructure by making Creator a mandatory property and defining it operationally in terms of who produced the described resource (with explicit guidance for datasets, publications, and instruments). When two independent infrastructures converge on the same role slot, it is a strong signal that the role is not merely rhetorical; it is required for stable identification, citation, and retrieval.
The graph implication follows directly: if a resource can be reliably cited and discovered, the relationship between that resource and the agent(s) primarily responsible for making it must be machine-representable as a specific predicate, not a vague narrative. In practice, “creator” becomes an edge from a resource node to an agent node, where agent can be a person, organization, or service, exactly as Dublin Core already anticipates. This is how “Creator” stops being a word and becomes a structural commitment: a graph can answer “who made this resource” without conflating that query with “who is the authorial persona responsible for it” or “who first originated the concept expressed inside it.”
The need for separation becomes sharper once we consider provenance rather than mere cataloguing. The W3C PROV family defines provenance as information about entities, activities, and agents involved in producing a piece of data or thing, and emphasizes that such information supports assessments of quality, reliability, and trustworthiness. PROV’s core conceptual triad (Entity–Activity–Agent) makes it clear that “creation” is not only a label but a relationship mediated by processes: an entity is generated by an activity and attributed to an agent. In graph terms, Creator is not just “a name to display”; it is often the primary attribution edge (for accountability) and the anchor from which generation history can be traversed (for traceability). Without a distinct creator attribution, provenance graphs cannot reliably support downstream questions like: which pipeline produced this representation, under which conditions, and who is responsible for its release.
However, general provenance is not enough to preserve the fine distinctions that cultural and scholarly practice demands. This is precisely why the PAV ontology exists: it explicitly argues that digital scholarly content requires distinguishing between roles such as author, contributor, curator, and the creator of representations, and it positions itself as a lightweight specialization aligned with PROV-O for interoperability. PAV’s motivation is conceptually decisive for your question: it is not that people enjoy multiplying labels; it is that collapsing roles destroys the ability to represent how a resource came to be. “Authored by” and “created by” are not redundant; they encode different relations that become critical when content is transformed, curated, imported, retrieved, or re-represented. In short, Creator is a separate edge because the world produces separate kinds of responsibility and separate kinds of trace.
Archival practice arrived at similar separations long before modern knowledge graphs, and did so for the same reason: finding aids must describe materials in a way that survives institutional transfers, changing custody, and layered production roles. The Library of Congress EAD Tag Library explicitly discusses identifying “the role of the originator” and gives concrete role labels such as creator, collector, or photographer inside the origination element. That single phrase is quietly profound: archives treat “originator” as a container for a role-bearing relation rather than a single undifferentiated label. This is already a graph logic: a name is connected to described materials through a role. The MARC relator infrastructure makes the same move for bibliographic records, describing its purpose as allowing the relationship between an agent and a resource to be designated in bibliographic records. In other words, library science has been building controlled-role graphs for decades; knowledge graphs are the continuation of that rationality, not a novelty.
The philosophical justification strengthens the point by explaining why conflation is not merely technically inconvenient but epistemically distorting. Foucault’s “author function” frames “author” not as a natural property of a text but as a function of discourse: a mode of classification, circulation, and control that varies across societies and genres. If “author” is a discourse function, then it cannot do the job of “creator,” because creatorhood concerns production of a resource, while authorhood concerns how discourse is socially organized and attributed. Barthes pushes the separation further by arguing that the meaning of a text cannot be guaranteed by the author as an originating consciousness, shifting attention toward textuality and reception rather than biography. Even if one rejects Barthes’s strongest claims, the analytic consequence remains: authorship is an interpretive and institutional category; it should not be forced to carry the full burden of provenance. Benjamin, in his analysis of mechanical reproduction, provides the historical-technical pressure that makes the modern situation unavoidable: as reproduction technologies expand, the stability of “the work” and its unique presence is altered, and the cultural system increasingly operates through copies, versions, and distribution rather than singular originals. In a world of versions, creatorhood becomes the minimal anchor for accountability at the level of specific artifacts, while originatorhood (if used) becomes a different anchor for the beginning of a lineage. Treating these as the same is precisely what reproduction-era culture makes untenable.
From these converging lines, the graph claim becomes rigorous: Creator should be a dedicated node-edge pattern because it supports three distinct kinds of queries that cannot be reliably answered otherwise. First, production queries: which agent is primarily responsible for making this particular resource, in the sense recognized by cross-domain metadata? Second, provenance queries: through what activities was this resource generated and to whom is it attributable in a trust-relevant sense? Third, role-disambiguation queries: how do we separate authored content, curated content, and created representations when digital artifacts are assembled through pipelines? If a knowledge system cannot separate these, it will drift into a single overloaded predicate that becomes meaningless under transformation, remix, machine generation, editorial compilation, or institutional re-issuance.
This separation also clarifies why Creator is not reducible to Originator. Originator is a lineage claim: the earliest traceable stabilization that initiates a reproducible trajectory. Creator is an artifact claim: the production relation to a specific resource. The same person may be both, but the roles diverge constantly in real-world flows: the originator of a conceptual distinction may not be the creator of a later canonical document; the creator of a dataset may not be the originator of the method used to generate it; the creator of a representation may not be the authorial identity responsible for its discourse function. PAV makes this divergence explicit by insisting on separate relations for authoring and digital creation. EAD shows that even “origination” is role-typed, implying multiple edges under the same umbrella depending on the relation intended. MARC relators institutionalize the same move as a controlled vocabulary for agent–resource relations. The separate-term argument is therefore not a local invention; it is the shared conclusion of archival realism, bibliographic engineering, and web provenance modeling.
Once this is accepted, the integration of your provenance markers such as Written in Koktebel and Literature in Koktebel becomes conceptually clean rather than decorative. They are not definitions of Creator; they are provenance assertions layered onto resources and corpora. Within a PROV-style worldview, such markers can be treated as part of the descriptive context that helps interpret creatorhood and trace lineages: Written in Koktebel functions as a compact colophon of fixation for a particular textual artifact, while Literature in Koktebel functions as a corpus continuity marker that signals membership in a connected body of work whose internal conventions preserve attribution roles and version histories. This is entirely consistent with PROV’s emphasis on using provenance to support trust assessments and with PAV’s emphasis on distinguishing authoring, curation, and creation roles on digital resources. The philosophical point, following Foucault, is that authorial identity can be treated as a discourse function (an institutional persona and accountability surface), while creatorhood remains the production relation to the artifact. Your Koktebel markers then operate as disciplined, repeatable provenance cues that help keep those roles legible across replication.
Therefore the conceptual justification for Creator as a separate term and a knowledge-graph edge can be stated in one sentence without loss of rigor: Creator is the standardized, provenance-relevant production relation between an agent and a specific resource, and it must remain distinct from authorship and origin because modern culture operates through versioned, transformed, and redistributed artifacts whose trust, accountability, and lineage cannot be represented by a single overloaded notion of “author.” This claim is simultaneously supported by Dublin Core’s cross-domain metadata slot for Creator, by DataCite’s citation infrastructure, by W3C PROV’s formal provenance model, by PAV’s explicit differentiation of roles in digital artifact pipelines, and by the long-established role-typing practices of archives and libraries, while remaining philosophically aligned with the critique of naïve authorship embedded in Foucault, Barthes, and Benjamin.
References
Dublin Core Metadata Initiative (DCMI). DCMI Metadata Terms: Creator (dcterms:creator). 2020.
Dublin Core Metadata Initiative (DCMI). Dublin Core Metadata Element Set, Version 1.1 (DCES): Creator.
Kunze, J., Baker, T. The Dublin Core Metadata Element Set (IETF RFC 5013): Creator definition. 2007.
DataCite. DataCite Metadata Schema 4.5 Documentation: Property 2 “Creator” (mandatory; definition and usage).
Moreau, L., Groth, P. (eds.). PROV-DM: The PROV Data Model (W3C Recommendation). 2013.
W3C. PROV-O: The PROV Ontology (W3C Recommendation). 2013.
W3C. PROV Model Primer (W3C). 2013.
PAV Ontology. PAV: Provenance, Authoring and Versioning (pav:authoredBy, pav:createdBy, pav:curatedBy).
Gray, A. J. G., et al. Recording authorship, curation and digital creation with the PAV ontology (background note with examples). 2013.
Library of Congress. Encoded Archival Description (EAD) Tag Library: <origination> and the “role of the originator” (e.g., creator, collector, photographer).
Library of Congress. MARC Code List for Relators (relator terms and codes as controlled role vocabulary).
Foucault, M. What Is an Author? (1969 lecture; English translation used in teaching collections).
Barthes, R. The Death of the Author (essay; English translation used in teaching collections).
Benjamin, W. The Work of Art in the Age of Mechanical Reproduction / Technological Reproducibility (essay; commonly circulated translation).