Beyond ethics and aesthetics: perceived lay theory violations and agency disruption in AI-generated art

The rapid integration of artificial intelligence into the realm of art creation has ignited a fervent debate, spanning from profound fascination to outright resistance. At its core, this technological advancement compels a fundamental re-examination of what constitutes creativity, artistic identity, and the very essence of human artistic endeavor. Research indicates that the devaluation of AI-generated art is not merely a matter of aesthetic preference but stems, in part, from a disruption of deeply ingrained cultural understandings of creativity. These intuitive frameworks often tie artistic value to perceived human creative performance, particularly the elusive "spark" of embodied insight and a coherent history of human agency. Unlike traditional artistic processes, AI is perceived as proceduralizing the very associative processes that underpin creative breakthroughs, generating artifacts where the link between the act of creation and the final outcome is no longer readily discernible. This creates a psychological chasm: observers may acknowledge the aesthetic merit of an AI-generated piece yet struggle to anchor it to a recognizable human generative process.

Drawing on frameworks of agency mapping, artistic communication, and appraisal-based emotion, recent analyses reveal how AI disrupts the mechanisms audiences typically employ to value and categorize art. These mechanisms, rather than focusing solely on the artifact’s surface cues, depend on reconstructing a human creative performance that links intention, process, and outcome into a unified generative history. When this reconstruction falters, artifacts fail to satisfy established psychological frameworks such as magical contagion, the extended self, the artistic design stance, and communicative inference, all of which rely on the perception of a human mind as the originating source of creative action. This structural breakdown elicits two core emotional responses: aesthetic anger, triggered by violations of expectations regarding human creative authorship, and ontological unease, arising from the simulation of uniquely human creative processes by non-biological systems. The ongoing discourse surrounding AI-generated art also highlights organizational framings—AI as Tool, Collaborator, or Threat—which attempt to stabilize the uncertainty inherent in agency reconstruction. Understanding these cognitive and emotional dynamics is crucial for clarifying why AI-generated art provokes resistance, even when its outputs are aesthetically compelling.

The Shifting Landscape of Artistic Creation

The emergence of AI-generated art has presented a significant challenge to deeply held cultural assumptions about the value and definition of art. While practical concerns regarding appropriation, job displacement, and copyright ambiguity remain prominent, they do not fully encompass the spectrum of public reaction. Responses range from intense curiosity and fascination to persistent resistance towards machine-mediated creativity. This friction extends beyond mere technical or legal debates, probing fundamental questions about creativity, authorship, and the very categorization of art.

In recent years, generative processes have achieved a level of sophistication that allows them to mirror human creative outputs across various domains. In music, researchers have demonstrated high-fidelity audio synthesis capable of replicating complex harmonic structures. Similarly, in literature, observers have struggled to distinguish machine-made poetry from human-written verse. The visual arts have seen particularly salient trends, with diffusion architectures producing high-fidelity synthesis that replicates stylistic patterns eliciting evaluative judgments comparable to human art. Despite these perceptual similarities, a consistent gap has emerged between sensory appreciation and perceived value. The knowledge of AI authorship demonstrably alters judgments of creativity, authenticity, and artistic merit, even when observers evaluate identical artifacts. In some instances, observers have even preferred AI-generated works when their origin was unknown, yet exhibited a subtle negativity bias when works were labeled as AI-generated. Identical artifacts have been rated as less creative, intentional, and authentic when attributed to machines rather than humans—a phenomenon described as the "creator label effect." This bias persists even when no perceptible differences exist between artifacts.

The Psychological Underpinnings of Artistic Value

These findings suggest that resistance to AI-generated art is not primarily a reaction to its aesthetic quality but rather to its perceived origin. Such responses may reflect broader anthropocentric motivations to preserve the distinctiveness of human creativity. When observers cannot readily map an artifact onto a coherent model of human generative activity, evaluations shift, often resulting in devaluation and resistance. This resistance aligns with interpretations that locate the effect primarily in higher-level interpretive systems concerned with authorship and meaning, rather than in basic sensory-motor aesthetic processing.

The present analysis organizes responses to AI-generated art into three broad framings reflected in prior scholarship: AI as a tool extending human creativity and agency; AI as a collaborator participating in a shared creative process while preserving human agency; and AI as a threat, wherein generative systems are described as disrupting human agency and creative labor. These framings shape how audiences interpret AI-mediated artworks and engage with lay theories of creativity and the psychological processes through which artistic value is typically inferred. Depending on which framing is activated, audiences may accept outputs as extensions of human agency or devalue them as violations of expectations surrounding authorship and intentionality.

Attribution-based theories of emotion provide a framework for understanding the affective consequences of these shifts. Within these models, emotional responses arise from appraisals of causal structure, including locus (internal vs. external), stability (temporary vs. enduring), controllability (manageable vs. uncontrollable), and perceived intentionality. When AI systems produce outputs that appear to bypass human generative processes, the inferred locus of causation may shift away from the human agent, perceived controllability may decrease, and the generative process may appear stable and replicable. Such appraisals may be associated with affective responses consistent with anger, reflecting perceived violations of norms surrounding authorship and the creative process. This shift may also contribute to a form of ontological unease when observers confront the possibility that algorithmic systems can produce artifacts resembling human expression without corresponding intentionality or internal states.

Together, these attributional and affective dynamics help explain the consistent devaluation of AI-generated outputs reported in prior work, as well as reduced willingness to attribute authorship or fully recognize such outputs as art. Rather than reflecting differences in perceptual quality, these responses suggest that AI-generated outputs may be experienced as destabilizing the boundaries of what counts as art.

Lay Theories of Creativity and the “Spark”

Beyond formal aesthetic analysis, audience responses to art are profoundly shaped by culturally ingrained lay theories of creativity. These implicit frameworks structure expectations regarding how creative works should originate and how inferred creative processes contribute to judgments of authenticity. Cultural narratives surrounding creativity often portray artistic insight as mysterious, ineffable, or inspiration-driven. This reliance on process-based cues is evidenced by the effort heuristic, in which perceived labor systematically influences judgments of quality. Such aesthetic-experiential frameworks extend even to highly technical domains, where intuitive and embodied judgment continues to shape evaluation. In the context of computational creativity, these findings highlight how judgments of value depend on an inferred history of production rather than surface features alone.

Lay theories function as essential, albeit technically incomplete, belief systems that allow individuals to navigate and attribute meaning to complex phenomena. In the domain of aesthetics, these theories prioritize the presence of a "human trace"—inferred intention, effort, or authorship—as a central basis for value, a concern that is also reflected in philosophical accounts of authenticity and artistic origin. Absorbed through cultural narratives rather than explicit instruction, these frameworks portray artistic creation as arising from more than technical skill or deliberate effort. Central to this belief structure is the expectation of a creative "spark"—an ineffable quality associated with inspiration, insight, and authentic expression, and the idea of innate genius. Rather than isolated beliefs, these lay theories operate as an integrated system governing expectations for both creation and reception.

Although creativity is actively being redefined to accommodate AI-generated processes, evaluations of creative work by lay audiences remain shaped by lay theories of creativity and the psychological mechanisms used to infer human creative performance, which privilege embodied effort, nonlinear insight, and human agency. When AI-mediated processes appear to bypass these expected pathways—through the proceduralization of generative activity, curatorial workflows, or the perceived outsourcing of insight generation—such shifts are likely to be interpreted as a violation of lay theories of creativity and authorship, not merely as technical innovation. This perceived disruption may alter how the artifact is mapped onto human agency, shaping appraisals of external causation, reduced controllability, and high replicability. These appraisals, consistent with attribution-based models of emotion, are theorized to give rise to anger and ontological unease in responses to generative technologies.

The Demystification of the “Aha!” Moment

A central element of lay theories of creativity is the "myth of inspiration," often imagined as a sudden "Aha!" moment in which an external muse bestows artistic insight. The term "inspire," derived from the Latin "to breathe into," reflects early mystical conceptions of creativity as a divine gift. During the Romantic era, as theorists reconceptualized this "breath" as an internal faculty, the spark of creation evolved from a divine infusion into a marker of innate genius. Consequently, the individual creator became the privileged originating source of the work. By relocating this creative impetus to the human psyche, the Romantic framework redefined artistic output as an index of human essence—a belief that continues to underpin modern expectations of authenticity. This legacy helps explain why algorithmic generation is experienced not merely as a technical shift, but as a challenge to foundational assumptions regarding creativity and authorship; in this view, AI acts as an implicit violation of long-standing lay theories.

Generative AI may disrupt this narrative of inspiration by altering how effort and intentionality are mapped onto artifacts. Even when AI-generated works satisfy formal product-based criteria for creativity—novelty, utility, and surprise—audiences may experience a disconnect between the artifact and the expected human generative process. According to the effort heuristic, perceived labor functions as a proxy for artistic quality. Automated generation complicates the inference that creative output reflects effortful human cognition.

AI systems instantiate generative processes that resemble formal accounts of associative creativity. By producing outputs through large-scale probabilistic recombination, these systems foreground the possibility that creative outcomes arise from structured search across patterns of association in representational space—a principle consistent with associative theories of creativity and developed in contemporary accounts of combinatorial and search-based models of creative cognition. While observers may recognize that the human retains curatorial oversight, the AI contributes outputs that cannot be fully anticipated. This unpredictability can function as a source of emergent surprise, allowing the collaborator framing to treat this distance between intention and outcome not as a loss of control, but as the site of generative emergence itself. Crucially, however, this generation of autonomous surprise simultaneously demystifies the traditional "aha" moment, effectively outsourcing the "magic" of the creative performance to an algorithmic process.

In this sense, AI does not eliminate the "aha!" moment but repositions it as an emergent property of distributed, non-local recombinatory processes rather than a discrete event within an embodied human mind. When algorithmic systems produce outputs that resemble the results of human insight, the inference that such outputs originate from uniquely human processes becomes less stable. Because lay theories of creativity depend on inferences about agency, effort, and intentional transformation, attributions of "magic" are weakened when these generative processes are externally visible or plausibly mechanized. What is experienced as magical is thus reinterpreted as the product of structured generative procedures rather than ineffable insight.

Disembodied Creativity and the Prompter’s Role

Demystification reshapes perceptions of the human prompter. Within the "genius/spark" framework, artistic value has been tied to creative performance—the visible, embodied transformation of an idea into material form. Aesthetic judgment tracks the observable "action" of creation: traces of problem-solving and iterative engagement with a medium. When AI serves as the primary generator, the prompter’s role shifts from execution to orchestration. This is characterized as "disembodied creativity," in which the creator is partially detached from the material feedback loop. Without direct causal interaction with the medium, the work may fail the "Lay Expectation of Evident Effort"—the belief that authentic artistic value requires visible struggle.

AI shifts the locus of creative cognition from associative, divergent recombination to collaborative navigation of latent space. Agency is no longer traced in embodied struggle or iterative negotiation with material resistance, but in high-level selection, semantic framing, and evaluative decision-making. While traditional media provided material resistance—the technical and conceptual demands that require creators to negotiate intent with a physical medium—AI interaction replaces this material negotiation with semantic negotiation. Semantic negotiation refers to the meta-cognitive and symbolic effort involved when a prompter navigates the probabilistic constraints of a generative model to achieve a specific intentional outcome. In this framework, the prompter does not bypass cognitive labor; rather, they refine intent through iterative engagement with the model’s outputs—a process characterized as a recursive feedback loop. Expert prompting, in this sense, constitutes a specialized, learned form of human skill, shaping coherence and meaning even as the associative work is performed by the model.

The audience’s struggle to recognize human effort in AI art is often reinforced by the "Romance of Computation," a framework in which there is a tendency to attribute autonomous creative agency to machines. As observed, this narrative often obscures the role of human input by leading observers to overlook the prompter’s generative contributions. Consequently, while the prompter remains the locus of initial generative direction—structuring the iterative prompt-system interaction that produces the artifact—the observer does not attribute the resulting output to a coherent human creative performance. Instead, the prompter is often reclassified as an evaluative curator rather than a generative agent. This shift weakens the perceived link between artifact and human creative agency and ultimately disrupts the process of agency mapping.

Within expert practice, prompting involves nuanced directive and interpretive control over model outputs—a process characterized as highly specialized—but this form of agency is not readily legible as part of the artifact’s perceived creative origins. Importantly, observers rely on culturally ingrained assumptions that art should embody a trace of human creative performance. In this context, the "effort heuristic" functions as a primary cue supporting the attribution of human agency; however, effort alone is not sufficient. When generative processes are perceived as procedurally mediated rather than grounded in human action, the artifact is less readily imbued with the properties of authorship and intentionality that underpin lay theories of creativity.

In sum, AI art prompts a renegotiation of lay theories along multiple dimensions: it relocates the moment of insight from the spontaneous "spark" to iterative curation, reconfigures the embodied linkage between artist and artifact, and transforms cognitive labor into meta-cognitive, strategic orchestration. These shifts are emotionally charged: the move from effortful "struggle" to high-level "semantic navigation" results in negative appraisals, including ontological unease and aesthetic anger, reflecting friction between human-centric definitions of art and emerging collaborative paradigms.

Agency Mapping and the Psychological Foundations of Artistic Value

The human-centric schema of art—rooted in Romantic-era conceptions of creativity—assumes that artworks are the expression of a singular human creative performance with a traceable generative source. Audiences use this human-centric schema to inform a coordinated set of psychological mechanisms through which observers infer meaning, intention, and value. These mechanisms—including the concept of the extended self, agency detection, the artistic design stance, and the role of magical contagion—function together as an agency mapping system. This system, which also incorporates principles of artistic communication and perceived scarcity, anchors the artwork to a perceived human source. When AI disrupts this system, it produces predictable emotional consequences: when observers cannot reconstruct a human generative pathway, they often experience "aesthetic anger" at the violation of creative norms and "ontological unease" regarding creations unanchored in human cognition. These responses arise not from superficial dislike, but from the perceived breakdown of mechanisms that normally anchor meaning and social grounding to human authorship.

Artistic value, in this sense, is not derived from the artifact alone but from the observer’s ability to map it onto a plausible history of human cognition and embodied performance. When this mapping succeeds, the work is experienced as an expression of a mind; when it fails, the artifact risks appearing procedural, ungrounded, or socially inert. AI-generated art is uniquely positioned to disrupt this system because it obscures or replaces the cues that these mechanisms rely on to anchor human agency: the locus of authorship shifts away from a singular human source, the human mind behind the generative process becomes displaced, and effort is no longer reconstructible. Consequently, attribution processes are strained across multiple cognitive levels simultaneously.

This section examines how each mechanism contributes to agency mapping and why its disruption leads to systematic devaluation, resistance, anger, and ontological unease. Rather than reflecting a single deficit, negative responses to AI-generated art emerge from a cumulative failure to sustain the perception of a human-originated creative act. Importantly, these mechanisms operate on the presupposition of not just agency, but specifically human agency—an assumption that becomes unstable under AI-mediated production.

Scarcity, Effort, and the “Aura”

Artistic value is related to scarcity in the sense that a work is implicitly believed to be the singular outcome of a unique creative act. Walter Benjamin proposed the concept of the "aura"—the perceived authenticity and unique presence of a work in time and space—to highlight how originality and uniqueness anchor perceptions of authenticity. Psychologically, audiences rely on scarcity as an effort heuristic: rare works signal that the artist’s performance was singular, irreplaceable, and required substantial effort. AI art undermines this heuristic through its capacity to generate countless high-fidelity variations with negligible friction. Consequently, the "aura" collapses, a shift that destabilizes traditional valuation systems and weakens the observer’s ability to reconstruct human effort, thereby reinforcing the perception that AI art lacks intrinsic worth.

The Extended Self and the Role of Tools

The perceived erosion of the aura reflects a disruption of the "extended self." Michael Belk proposed that material objects can become incorporated into a person’s identity, functioning as symbolic extensions of the self. Building on this framework, scholars argued that artworks serve as paradigmatic extended-self objects, insofar as they are interpreted as repositories of a creator’s identity, effort, and embodied performance.

In artistic contexts, this sense of self-extension is supported by the perceived connection between the artist and their tools. As argued, the value of an artwork is closely tied to the physical process of its creation; instruments such as brushes and chisels serve as conduits through which human intention is manifested in material form. When the creative process is perceived as transparent and the tool as continuous with bodily action, observers are more readily able to interpret the resulting artifact as a material continuation of the creator’s embodied agency.

AI systems, however, are less often experienced as transparent tools. Instead, observers frequently perceive them as semi-autonomous systems operating according to opaque generative processes. Unlike traditional artistic instruments, which recede during skilled use, AI systems tend to remain experientially salient as independent sources of generative output. This perceived opacity is associated with attributions of partial autonomy or system-level agency, which in turn can decouple the human creator from the final artifact.

Consistent with this shift in perceived agency structure, observers are more likely to attribute reduced intentionality and creativity to AI-generated works relative to human-made ones. Importantly, this effect persists even when human involvement is present, suggesting that what changes is not the actual causal role of the human, but the perceptual accessibility of that role within the artifact. As a result, AI-generated outputs are less likely to be experienced as direct extensions of the creator’s self, even when they originate within human-guided workflows.

Intention, Communication, and the Human Touch

Bullot and Reber argued that audiences adopt an artistic design stance—a default orientation toward artworks as intentional acts. Viewers interpret brushstrokes, composition, and style as deliberate choices that express meaning, an interpretive framework that presupposes a conscious agent. Humans are also predisposed to perceive intention in meaningful patterns through an agency-detection bias, which inclines audiences to assume that artworks originate from conscious, purposeful agents. Together, these mechanisms allow viewers to experience artistic features as traces of intentional human activity embedded within the artifact itself.

AI-generated art disrupts both of these cognitive tendencies. Although audiences may appreciate the surface aesthetics of AI-generated works, the perceived absence of a subjective mind or lived experience can make it more difficult to attribute authentic intention to the artifact. For example, studies found that participants judged robot-produced paintings as art to a similar extent as human-produced ones, yet were significantly less willing to consider the robots themselves "artists," reflecting reduced attributions of artistic intention. This phenomenon has been described as "unintentional intentionality," in which a machine produces the appearance of purposeful design without an underlying conscious source. In this sense, AI-generated works may partially satisfy the perceptual features associated with the artistic design stance while weakening the inference that those features originate from a situated human subject.

Magical contagion—the belief that physical contact or proximity transfers an invisible essence from person to object—operates as a psychological mechanism that links the artist’s embodied performance, intentionality, and the aura of the extended self to the artifact. Scholars demonstrated that the value of an original artwork lies not in perceptual features alone, but in the belief that the artist’s physical presence confers authenticity. Within this framework, perceived effort functions as a diagnostic cue: the "effort heuristic" allows observers to infer the depth of creative essence transferred, connecting the human mind, body, and intentional action to the artifact itself. When this chain is broken—as in AI-generated art, the artifact loses the aura of the extended self and the perceivable trace of human intention, undermining its perceived authenticity and value.

This process is further mediated by sensitivity to "human marks"—idiosyncratic traces of manual execution conveying motor origins. Studies found that perceptual features tied to natural human movement—such as the dynamics and timing of strokes—enhance aesthetic engagement. Importantly, these human marks serve not merely as signs of labor, but as cues that an originating insight occurred; through these traces, observers perceive authorship as transferring to the artifact. AI-generated art disrupts this transmission. Although a human may provide prompts, a non-biological system executes the transformation, disrupting the extended self-mechanism. Without effort cues or subtle motor signatures, the conduit between creator and artifact attenuates, and the human prompter’s agency becomes difficult for observers to register, weakening the symbolic link required for inter-agentic communication.

Art is not only valued as intentional but also as communicative in a qualified sense. Researchers did not posit a direct transmission of meaning from artist to viewer. Rather, they conceived of artworks as containers of embodied constraints that structure interpretation, allowing viewers to infer the presence of an originating perspective. In this sense, the artwork functions as a bridge—not between fixed meanings, but between a creator’s situated engagement with experience and the viewer’s reconstructive, interpretive activity.

Appreciation is enhanced when viewers can construct a narrative or perceived "aboutness" behind a work, an inference more readily anchored to human creators. Fundamentally, this process depends on the assumption that the artifact is the product of a mind engaged in the world. When this assumption weakens—as in AI-generated art, where the generative pathway bypasses or obscures a traceable human source—the viewer’s ability to sustain this communicative inference is disrupted. Rather than failing to decode a message, the viewer struggles to justify the presence of a communicative act at all, and the work may be experienced as a socially sterile or indeterminate simulation.

Organizational Frames and Emotional Responses

As established, artistic valuation depends on the observer’s ability to map an artifact onto a reconstructible history of human generative activity. When this mapping becomes uncertain or breaks down, observers do not respond uniformly. Instead, they adopt interpretive framings that attempt to stabilize or account for the disrupted relationship between artifact and human agency.

Discourse around AI-generated art organizes into three dominant framings: AI as tool, AI as collaborator, and AI as threat. These framings are not merely descriptive; they reflect different ways of interpreting how AI-mediated processes engage—or fail to engage—the lay theories and psychological mechanisms that underpin artistic value. Through tool and collaborator framings, observers preserve or redistribute their perception of human creative involvement, allowing observers to maintain partial alignment with expectations of effort, agency, intention, contagion, and communicative intent. However, when the perception of these cues is sufficiently bypassed—when effort flattens, agency redistributes, and the artifact is experienced primarily as the product of a procedural system rather than a human generative path—the mapping collapses and the work is more likely to be interpreted through a threat framing.

The emotional responses associated with AI-generated art follow from these interpretive outcomes. Aesthetic anger emerges when the observer appraises the breakdown as a violation of the norms governing human creativity—specifically intention, effort, authorship, or creative performance. Ontological unease arises when the breakdown destabilizes the category itself, calling into question the fundamental assumption that creative artifacts are expressions of human minds.

The Tool Framing: Preserving Human Agency

In the tool framing, AI is interpreted as an extension of human agency, analogous to traditional creative instruments. This framing relies on the extended self hypothesis; in their respective formulations, scholars posited that tools can function as transparent conduits of embodied human intention and creative performance. When observers perceive the technology as a passive instrument, the human prompter retains authorship, and the AI output is experienced as a material instantiation of human insight rather than the product of a semi-autonomous system. For this framing to remain stable, the audience must successfully employ the artistic design stance, decoding the human’s "why" behind each formal choice to recognize intentionality and expressive meaning.

Human valuation of creative products is deeply tied to magical contagion, the belief that physical contact or proximity transfers an enduring essence from person to object. In traditional artistic practice, instruments function as vessels for this contagion: brushes, pens, and chisels channel the artist’s internal "spark" into material form, leaving behind perceptible human marks—subtle traces of motor execution, dynamics, and timing of strokes—that audiences interpret as evidence of effort and creative insight. This externalization is reinforced by the effort heuristic, whereby observers use visible struggle and time investment as proxies for the depth of the creative essence transferred. Studies demonstrated that audiences respond positively to perceptual cues of human motor activity—such as brushstroke dynamics or idiosyncratic marks—that signal an embodied, human origin. Furthermore, the knowledge that an artist labored extensively on a piece ("time presence") enhanced its perceived authenticity and value. In these cases, AI can be perceived as a transparent extension of human agency only if such cues remain legible and attributable to the human prompter.

Historically, similar legitimacy challenges arose with new technologies. When photography emerged, it was initially dismissed as a mechanical process lacking the "soul" of manual execution. The medium gained acceptance as a creative tool once audiences learned to map photographs back to the photographer’s embodied presence; framing, timing, and perspective became perceptible signatures of human intention, preserving the pathway for magical contagion. Likewise, a tool framing for AI requires that the observer recognize the human’s curatorial choices—the selection of prompts, guidance of outputs, and iterative refinement—as the locus of creativity.

AI disrupts the traditional boundary between creator and instrument, as the system’s high-level execution often appears to mimic the autonomous insight previously associated with the human mind. While the human prompter provides inspiration or constraints, the visible transformation occurs within an algorithmic system. Because the resulting artifact often lacks the idiosyncratic motor traces and visible effort cues that convey embodied agency, audiences may struggle to perceive the human "spark" as materially inscribed. This failure weakens the extension of the self and the symbolic and affective transmission that sustains inter-agentic communication. A perceptual gap is supported by previous research: audiences attributed less creativity and intentionality to AI-generated works than to human-made ones, even when humans were involved in curatorial or prompt-guiding roles.

Furthermore, AI’s procedural and semi-autonomous operation threatens the magical contagion pathway. Without perceivable human effort or struggle, observers default to attributing agency to the algorithm rather than to the human. Even if surface aesthetics are compelling, the work appears procedural rather than expressive; the human signature becomes illegible, and the artifact risks being interpreted as socially and emotionally detached. Maintaining a tool framing thus depends on strategies that preserve human legibility: emphasizing deliberate choices, iterative refinement, and curatorial oversight allows audiences to anchor agency and creative insight in the human, rather than the AI.

In sum, interpreting AI as a tool requires preserving or signaling the cues that link human intention, effort, and embodied execution to the final artifact. By maintaining these markers, observers can experience the output as an extension of the creator’s self, sustaining magical contagion, the extended self, and the artistic design stance. Failure to preserve these cues may collapse the tool framing, pushing audiences toward collaborator or threat interpretations, where the AI is perceived as autonomous, human intention becomes detached from the artifact, and perceived structural violations of creative expectations elicit aesthetic anger, ontological unease, or devaluation of the work.

The Collaborator Framing: Distributed Agency

Beyond the tool framing, the collaborator framing positions the algorithm not as a passive instrument but as a co-creative partner, with humans and machines jointly shaping the final artifact. Historically, collaboration with non-human generative systems—such as electronic synthesizers or algorithmic composition systems—was theorized as challenging traditional boundaries of creative agency. Within this partnership, AI functions as a generative medium where, as posited, operation occurs not through direct manual control, but through complex, data-driven transformations within a high-dimensional space where the mapping from human intention to final output is inherently non-linear and non-transparent.

Empirical research suggested that humans navigate this collaborative space through distinct interaction styles, often acting as high-level directors who shape, evaluate, and integrate machine outputs rather than specifying every detail of the generative process. In these settings, the human role shifts toward curating and validating outputs to ensure alignment with intended goals. Even when the system appears partially autonomous, the human remains the primary locus of intentional direction, accountability, and evaluative constraint. Consequently, agency is experienced as distributed: the human provides vision, selection, and interpretive grounding, while the system contributes stochastic variation and emergent novelty.

However, this attribution of shared agency remains fragile. Studies demonstrated that even when observers cannot reliably distinguish between human- and AI-generated outputs, revealing the machine’s role nevertheless produces systematic, source-based devaluation of the work. Observers may acknowledge the novelty or aesthetic richness of AI-assisted outputs while simultaneously questioning whether the human retains sufficient authorship to satisfy the artistic design stance. Because the mapping from human intention to final outcome remains partially opaque within AI-mediated systems, the source of novelty is no longer attributed solely to the human creator.

For some observers, this redistribution of generative agency produces a form of evaluative tension or ontological unease. The work may remain aesthetically compelling while simultaneously signaling the loss of a unique and unified human signature. Because AI systems are often perceived as lacking the "internal motivation" or "subjective experience" associated with authentic expression, the collaboration is experienced as fundamentally asymmetrical. The audience can appreciate the artifact aesthetically while struggling to reconstruct a coherent human generative history behind it.

This tension becomes especially pronounced when artworks are evaluated as communicative acts rather than merely formal objects. Art is valued not solely for aesthetic form, but as a bridge between the creator’s lived experience and the viewer’s interpretive world. When a generative AI contributes substantively to the production process, this communicative bridge weakens because the system is perceived as unable to embody or transmit subjective experience. Even when a human prompter provides substantial curatorial guidance, aspects of the artifact may appear unanchored from a coherent lived perspective. The audience may continue to enjoy the work aesthetically, yet the social and epistemic link to a conscious human agent—the foundation of shared artistic communication—becomes destabilized.

Consequently, the extended self relationship between creator and artifact also weakens, as the boundary between human and machine agency becomes increasingly blurred. The artifact risks being experienced less as a communicative expression of human subjectivity and more as a simulation of expressive form, signaling human insight only partially or inconsistently. This produces a sense of distributed or plural authorship that is difficult to anchor in a single intentional source, generating epistemic ambiguity in agency attribution. At this stage, responses may shift toward interpretive instability that can, under certain appraisals, transition into threat-based evaluations, generating both aesthetic interest and subtle discomfort.

The Threat Framing: Synthesis of Disruptions

The threat framing does not emerge from a single violated expectation, but from the convergence of disruptions across the psychological and cultural systems that jointly sustain artistic value. When effort, agency, insight, intentionality, communicative grounding, and scarcity cues are simultaneously destabilized, the artifact can no longer be mapped onto a coherent model of human creative production. This cumulative breakdown—rather than any individual violation—gives rise to aesthetic anger and ontological unease.

Creative artifacts are culturally understood not only as products, but as outcomes of a specific kind of human generative process, characterized by embodied effort, intentional transformation, and situated cognitive engagement. Observers therefore do not evaluate artworks solely by their formal properties; they evaluate them by reconstructing the inferred human pathway through which those properties were produced. When AI-mediated generation disrupts the availability or coherence of these process cues, the causal link between artifact and human creative activity becomes difficult to sustain.

This disruption produces a shift in evaluative inference: the artifact may remain formally intelligible as an artwork, yet the underlying generative process becomes ambiguous. Here, the distinction between type authenticity and token authenticity becomes important. AI systems often preserve type authenticity, in that they reproduce the recognizable stylistic and structural features of artistic categories. However, they may weaken token authenticity by obscuring the specific human cognitive and historical pathway through which the artifact was generated. Importantly, this is not a failure of appearance, but a failure of process traceability.

The resulting instability is best understood as a breakdown in process-based grounding of category membership. Within lay theories of creativity, "art" is partially defined by the presumed presence of effortful, intentional, and embodied human cognition. Within the agency mapping system, these same assumptions function as inferential cues that anchor meaning and authorship to a human source. When token authenticity cannot be reliably reconstructed, these cues no longer integrate into a coherent attribution of human creative agency.

In this sense, the threat framing is not produced by formal failure, but by the structural incongruity between high type authenticity and weakened token authenticity. When observers can no longer reconcile these dimensions, effort, agency, insight, intentionality, communicative grounding, and scarcity cues fail to cohere into a unified model of human creative production. The resulting percept is not simply that the artifact is "AI-generated," but that the underlying structure of creative causation has become opaque, decentered, and psychologically unmappable. It is this collapse of coherent attribution—rather than any single violated expectation—that gives rise to aesthetic anger and ontological unease.

Conclusion: Reconstructing Agency in the Age of AI Art

AI-generated art challenges the psychological and cultural frameworks through which creativity and its products are evaluated. Across analyses, responses to AI-mediated works are shaped less by surface aesthetic qualities than by how observers reconstruct underlying lay theories of authorship, intentionality, and embodied effort. When artworks are evaluated as the outcome of human intention and historically grounded processes of creation, AI outputs disrupt the cues used to infer these properties, complicating judgments of authenticity and meaning.

The empirical implications are clear: emotional responses to violations of lay theories of creativity, such as aesthetic anger and ontological unease, are expected when AI-generated artworks disrupt norms of effort, intentionality, and embodied insight. Future research should explore how AI-labeled works reliably elicit stronger anger and existential discomfort compared to identical human-attributed artifacts. Furthermore, disruptions in agency attribution are anticipated, leading to lower judgments of creativity, authenticity, and emotional engagement for AI-generated works, mediated by reduced perceived intentionality and connection to a human communicative source.

Interpretive framing plays a crucial role. When AI is perceived as a tool that transparently extends human agency, attributions of authorship are preserved. Collaborative framings lead to distributed agency, while threat framings, emphasizing AI autonomy, intensify negative emotional responses. Individual differences, particularly expertise, are also likely to moderate these responses, with domain experts potentially showing reduced devaluation due to a better understanding of prompting and curatorial skill.

Ultimately, AI-generated art does not merely introduce a new class of aesthetic objects; it exposes the inferential structure through which observers attribute creative agency to artifacts. Across tool, collaborator, and threat framings, the critical variable remains the observer’s ability to reconstruct a coherent human generative pathway from available cues. Resistance to AI art, therefore, reflects not simply a preference for traditional media, but a systemic disruption in the cognitive and cultural mechanisms that link observable features of production to inferred human intention and meaning. The emotional responses associated with these disruptions—including aesthetic anger and ontological unease—underscore that artistic value is fundamentally tied to the reconstructability of a coherent human generative pathway within the artifact. This ongoing dialogue between technology and human perception promises to redefine our understanding of creativity for generations to come.