- The paper introduces a product-first, context-sensitive criterion for creativity that replaces the traditional Intentional Agency Condition.
- Empirical evidence and linguistic trends show that generative AI can outperform humans on creativity tests and alter attributions of creative value.
- The framework mitigates bias by emphasizing consistent output of novel and valuable artifacts, expanding creative evaluation beyond anthropocentric models.
Rethinking Creativity: The Status of Intentional Agency in the Age of Generative AI
Background: Intentional Agency and the Standard Definition of Creativity
The prevailing theoretical landscape in creativity research has been dominated by the Standard Definition (SD), which identifies two necessary and jointly sufficient criteria for creativity: novelty and value. In response to apparent counterexamples—particularly cases of accidental or natural outputs being classified as creative—many theorists introduced the Intentional Agency Condition (IAC). The IAC stipulates that creativity necessarily involves the action of an intentional agent, conscious of goals and engaged in purposive activity.
This process-first approach (SD+) has been rationalized on both intuitive and functional grounds. Intuitively, attributions of creativity are seen as vehicles for praise and resource allocation, which presuppose agency; functionally, the IAC serves to prevent the misallocation of praise and resources by ensuring that only agents capable of consistent creative output are recognized. This is further reinforced by appeals to linguistic practice and counterfactual scenarios where creative ascriptions to non-agents are perceived as unnatural.
Generative AI and the Descriptive-Functional Crisis of the IAC
The emergence of generative AI—specifically LLMs and diffusion models capable of producing domain-validated, novel, and valuable outputs across textual, sonic, and visual modalities—poses critical descriptive and functional challenges to the IAC. Empirical evidence now demonstrates that generative AI:
- Regularly outperforms human benchmarks on standardized creativity tests.
- Produces non-accidental, domain-relevant artifacts on demand and at scale.
- Invokes a pragmatic socio-economic dynamic, as evidenced by the significant allocation of attention and capital toward generative tools and their outputs.
Furthermore, corpus linguistic analyses across published literature and media (Google Ngram, NOW corpus) reveal rapidly rising attributions of creativity to AI systems, reflecting a substantial drift in ordinary language use and public intuition. These data undermine the descriptive adequacy of the IAC, as the philosophical claim of a linguistic consensus against AI creativity no longer obtains.
Functionally, application of the IAC in the current landscape introduces clear maladaptive biases:
- It systematically devalues novel and valuable AI outputs due to producer-identity effects and effort heuristics, now well-documented in the empirical literature.
- It risks discouraging resource allocation toward highly productive generative technologies, impeding the exploitation of AI for further creative production.
- It perpetuates forms of "algorithm aversion," narrowing opportunities for creative expansion through anthropocentric constraints.
These points collectively challenge the continued viability of the IAC as a general condition for creativity under contemporary and forthcoming technological regimes.
Context-Sensitive Conception and the Consistency Requirement
Rejecting the IAC as a necessary, global constraint, the paper advances a product-first, context-sensitive conception of creativity. Here, the primary functional value of "creativity" attributions is to identify and endorse consistent sources of novel and valuable outputs, independent of their agential underpinnings.
The proposed New Standard Definition (NSD) is:
An object is 'creative' if it is (a) novel, (b) valuable, and (c) the product of a system that can consistently generate novel and valuable objects. A system is 'creative' if it can consistently generate such objects.
Crucially, this definition supplants the IAC with a consistency requirement. Consistency is context-sensitive: it is not reducible to simple statistical frequency, but rather to domain-relative grounds for expecting further valuable novelty from a source. This revision circumvents the accidental/serendipitous production problem without enforcing process-based restrictions incompatible with generative AI.
However, the proposal is explicitly pluralist and context-sensitive. The IAC retains functional value within specific domains where expressive authenticity, responsibility, or legal accountability are relevant—e.g., in the cognitive science of human creativity, legal attribution of creative acts, or genres demanding sincerity such as love letters and commemorative speech. In these contexts, intentional agency remains an indispensable criterion for creativity ascriptions.
Consequences and Theoretical-Methodological Implications
The endorsement of a context-sensitive, consistency-based conception of creativity has wide-ranging theoretical and practical implications:
- Normative demarcation of creativity: The field must replace universal process-first constraints with a domain-adaptive taxonomy, formally recognizing generative AI as creative in general but allowing for the targeted reintroduction of the IAC where warranted.
- Bias mitigation: By removing the IAC from general application, product evaluations become less susceptible to heuristics that systematically undervalue AI outputs, supporting more objective baseline assessments.
- Expanded creative landscape: Classifying non-agentive yet consistent generative processes (e.g., AI, evolutionary processes) as creative facilitates investment in understanding and leveraging their creative potential, promoting both theoretical insight and technological advancement.
- Recognition of expressive authenticity: The pluralist approach clarifies that expressive authenticity is a specialized constraint; it is not intrinsic to the general concept of creativity but rather a local value in certain domains (art, communication, jurisprudence).
- Philosophical and methodological reorientation: The paper exemplifies a conceptual engineering strategy, advocating for theorists to actively shape the extension and function of "creativity" in response to linguistic and functional evolution.
Future Directions
This framework sets the agenda for further specification of where and how the IAC should be applied. Demarcating the boundaries of local domains requiring process-first constraints is now an urgent research problem, as is empirical tracking of ongoing linguistic and evaluative shifts. Additionally, expanded qualitative and quantitative assessments of creativity in AI-generated outputs are needed to refine the operationalization of the consistency criterion and contextual fitting of agency constraints.
The theoretical implications also extend to AI development, as incentive structures and system evaluation will increasingly rely on product-centric rather than process-centric metrics. This reorientation enables AI research to focus directly on the reliability and value of outputs without dogmatic adherence to anthropocentric models of cognition.
Conclusion
This article rigorously demonstrates that the Intentional Agency Condition is obsolete as a universal constraint on ascriptions of creativity in light of generative AI. The authors replace it with a context-sensitive, product-first formulation where consistency in producing novel and valuable outputs is the primary criterion. The IAC is preserved only where local domain utility demands it, especially in cases of expressive authenticity and legal responsibility. This reconceptualization, grounded in both empirical linguistic trends and functionalist arguments, aligns creative evaluation with contemporary technological and social realities. Future work must focus on systematically delineating local criteria and further refining the operational landscape of creativity assessment in the age of AI.