Creative Capacity: generating useful novelty.

A logistics company wants to reduce failed home deliveries without adding drivers, widening delivery windows or lowering service standards. The first meeting fills quickly with familiar proposals: send more reminders, ask customers to confirm earlier, improve route planning, add another notification before arrival. Each suggestion is reasonable because each remains inside the way the problem has already been represented.

Then someone questions an assumption that had survived unnoticed. Why must every successful delivery end at the customer's front door? The question changes the space in which solutions can exist even before any solution has been chosen. Local handoff points, building-level arrangements, neighbourhood partnerships and flexible pickup options become thinkable. Some are weak, some collide with regulation or economics, and some contain elements worth developing. The creative work now extends into what each emerging possibility could become, whether its logic can be made clear enough to examine, how it might be shaped into a coherent concept and where that concept could create genuine value.

We define Creative Capacity as the capacity to generate new ideas and transform them into meaningful concepts. Within the Unified Skills Map, we organise this capacity through six interrelated dimensions: Opening the creative space, Generating diverse ideas, Exploring ideas' potential, Clarifying the idea, Shaping the concept, and Articulating value and applications. The academic literature provides strong support for several of these functions and convergent support for the rest, while offering little basis for treating them as six independent faculties or a fixed chronological sequence. Contemporary creativity research increasingly describes creative work through problem construction, generation, evaluation, refinement and contextual judgment alongside divergent thinking (Pinkow, 2023; Tolkamp et al., 2022).

That wider view matters because creativity is still frequently inferred from the easiest part of the process to observe. A person produces many ideas, a product receives a high originality rating, or a team launches something novel, and the outcome is treated as a direct readout of an underlying creative ability. Creative products depend on knowledge, motivation, context, assistance and judgment as well as generative processes, and the processes used to create them can differ substantially even when the final outputs look similar (Long and Wang, 2022; Guo et al., 2022).

The six dimensions help us examine the work that occurs between an inherited view of a situation and a concept developed enough to carry meaning. The six dimensions remain functions within one changing process in which a new idea can reopen the original frame, clarification can reveal another possibility, evaluation can trigger further generation, and considering an application can force the concept itself to change.

Creative work can begin before an idea exists

The delivery team initially accepts a familiar representation: a delivery is successful when a driver reaches a specific address and hands the parcel to the customer. Most early proposals remain variations inside that representation. Better notifications, tighter scheduling and improved routing may all be useful, yet none changes what counts as a delivery.

Opening the creative space concerns the capacity to loosen, vary or reconstruct the assumptions and representations that govern what seems possible. Creativity research gives this function a strong foundation through problem construction, reframing, representation change, fixation, analogy and constraint research. The relevant mechanism concerns how a situation is represented, because that representation changes what memory, associations and possible actions become available for subsequent search.

Recent field research on ill-defined problem framing makes this process visible. Pham, Magistretti and Dell'Era (2023) observed 72 participants across six design-thinking workshops and identified analogical, associative and abductive operations through which participants retrieved existing representations, challenged them and constructed new frames. Their work directly supports the function we describe by showing people actively reconstructing the representation from which creative search proceeds, while leaving open whether that function forms a generic psychometric factor.

Fixation research shows the same mechanism from the failure side. Existing examples, familiar solutions and apparently fixed constraints can narrow later generation by making one region of the search space disproportionately accessible. The influence can survive even when the example was intended as inspiration. Wadinambiarachchi et al. (2024) provide a contemporary demonstration in a controlled visual-ideation study: novice designers using an AI image generator produced fewer ideas, less variety and lower originality than a no-support baseline while showing greater fixation. Because the result comes from a particular task and interface, it supports a narrower conclusion about how quickly external representations can anchor what creators subsequently explore.

Damadzic et al.'s (2022) meta-analysis and subsequent work by Tromp (2023) and Cromwell (2024) show that the relationship between creativity and constraint depends on the kind of restriction and how it is configured. Some constraints make the problem tractable, direct attention toward consequential trade-offs or force recombination away from obvious options. Others become destructive because they are excessive, rigid, contradictory or accepted as fixed when they are actually negotiable. Creative opening therefore includes distinguishing the conditions that genuinely define the problem from assumptions that merely arrived with it.

Analytical Capacity becomes closely involved at this point because analytical framing is concerned with constructing an adequate representation for understanding or investigating a question. Creative opening deliberately stretches or transforms the representation so that alternatives absent from the current frame become conceivable. The same operation, such as analogy or abstraction, may contribute to both. Its function inside the wider activity determines whether it is helping to establish an evidentially adequate frame or opening a new possibility space.

Generating diverse ideas creates material for development

Once a space has opened, the creative task often requires variation. Generating diverse ideas concerns producing multiple possibilities that differ meaningfully in direction, mechanism, structure or governing assumption. This is the most extensively studied part of Creative Capacity because divergent-thinking research has spent decades measuring ideational fluency, flexibility and originality.

That evidence is valuable precisely because it also reveals the limits of simple idea counts. Producing more responses increases the number of opportunities for an unusual response to appear. Forthmann (2021) describes the resulting fluency contamination problem: some originality scores rise partly because the person generated more material, not because the average quality or distinctiveness of that material improved. Contemporary measurement therefore increasingly separates response quantity from average, maximum or selected originality and treats semantic distance as one indicator rather than a complete measure of creative quality (Beaty et al., 2022; Weiss et al., 2021).

Ten cosmetic variations of the same delivery reminder may produce a high response count while leaving the search space almost unchanged, so meaningful diversity has to be judged by how far the underlying direction, mechanism or structure actually changes. Meaningful diversity can appear through a different mechanism, user, location, sequence, ownership model, assumption or underlying structure. A handoff network involving neighbourhood businesses represents a different direction from a better notification system because it changes how successful delivery is accomplished rather than modifying the message sent before the same process.

Serial-order research helps explain why continued generation can matter. Easily accessible associations tend to appear early, while later search may require greater persistence or a shift in direction. Ahn, Park and Moulder (2023) show that the temporal order of ideas changes through the process and can be affected by social communication. Earlier work on the serial-order effect similarly indicates that later responses can become more original as creators move beyond readily available associations (Beaty and Silvia, 2012). Persistence and flexibility provide different routes, because one continues searching within a direction while the other moves to another region of the possibility space.

Quantity also carries an opportunity cost because time spent creating another thirty possibilities cannot be used simultaneously to develop the strongest three. Highly remote ideas can increase novelty while reducing feasibility or contextual fit, and a large unstructured option set creates its own evaluation burden. Gillier and Bayus (2022), studying professional creative groups, found that building on novel ideas could increase later novelty while lowering feasibility. Generation therefore creates material for creative development without guaranteeing that the material is useful.

Group ideation introduces a different set of conditions: interaction can stimulate associations that individuals would not retrieve independently, while production blocking, evaluation apprehension and conformity can suppress contributions. The established brainstorming literature has documented these losses for decades, and recent work suggests that hybrid arrangements, such as individual generation followed by interaction, can mitigate some costs under suitable conditions (Diehl and Stroebe, 1987; Ostrowski, Woolley and Haan, 2022). A collaborative setting changes the conditions under which individual Creative Capacity is expressed while leaving the underlying individual capacity conceptually distinct.

The first idea is often only a starting material

Creativity becomes more interesting once an idea exists. A raw possibility is rarely a miniature version of the final concept waiting to be polished. It can be incomplete, ambiguous or internally inconsistent while still containing something worth pursuing. Exploring ideas' potential concerns using that emerging idea as a new generative starting point and examining what it could become, where it could lead, what other possibilities it creates and how it might behave in another context.

The exact label is ours and has no established psychometric construct carrying the same name. Its academic support comes from several neighbouring traditions, including the Geneplore model, elaboration, iteration, simulation, building on ideas, analogical extension and design exploration. Finke, Ward and Smith's (1992) Geneplore framework remains particularly relevant because it treats generated structures as objects that can subsequently be interpreted and transformed, rather than treating generation as the endpoint of creative cognition. More recent integrative work describes the relationship between generative and evaluative processes as increasingly fluid and recursive (Pinkow, 2023).

We distinguish generating from exploring by function, while accepting substantial overlap in practice. Generation begins with the wider task or possibility space and creates candidates. Exploration begins with an emerging candidate and asks what new trajectories, combinations or implications can be created from it. The two can overlap because exploring one possibility may produce several new ideas, and an unexpected new idea may become the object of further exploration.

Gillier and Bayus (2022) provide useful evidence from nine professional creative groups moving from earlier ideas towards final concepts, where ideas used as stimuli were disproportionately likely to enter the final concept. Building on an idea was not inherently beneficial, however. The effect depended on which idea was developed, and highly novel stimuli could generate trade-offs with feasibility. Exploration therefore changes the trajectory of a concept without providing an automatic improvement mechanism.

Iteration research adds a further qualification through Taranu et al. (2022), who found that repeated iterations in a LEGO creative task increased originality while reducing task appropriateness. The finding matters because it separates development from progress on every criterion. An idea can become more unusual and less suitable at the same time. Exploring potential opens consequences and alternatives, some of which eventually strengthen the concept and others of which reveal that the direction should be abandoned.

Physical representations can also widen exploration by making functions and analogies easier to inspect. Gomes et al. (2022), studying conceptual product design, found that physical models supported understanding of analogical sources and contributed to richer functional ideation. The representation participated in the creative process by supporting the discovery of further possibilities instead of serving only as a record produced after thought had finished.

The evidence therefore supports Exploring ideas' potential as a real creative function while leaving its status as an independent faculty unresolved. In practice, some episodes may contain very little of it, while others may spend most of their creative effort developing the consequences of one promising seed.

An idea has to become intelligible before it can become a concept

The later dimensions move into territory that mainstream creativity measurement studies less directly. Divergent-thinking tasks can count responses and product ratings can judge finished outputs, while the transition from an initially vague possibility to an intelligible concept is harder to isolate. Our framework therefore draws more heavily here on convergent evidence from creativity, design cognition and external-representation research.

Clarifying the idea concerns making an emerging possibility sufficiently explicit for its central meaning, intended mechanism, boundaries and unresolved ambiguities to be identified. The evidence for this function is convergent, drawing on several neighbouring constructs instead of a mature clarification construct. Research on elaboration, externalisation, explanation, sketching, physical models and representational refinement nevertheless gives the function a credible basis.

A creator can have a useful intuition without yet being able to state exactly what it means. The neighbourhood handoff idea might initially exist as the vague instruction “use local shops somehow.” That phrase can feel concrete because the objects are familiar, while almost every consequential part remains unspecified. Who receives the parcel, how custody transfers, which locations qualify, how customers select the option, how failed handoffs are handled and what the model changes for drivers are still unresolved.

A sketch, short explanation, service map, model or conversation can expose ambiguity by making missing relationships visible and forcing the creator to distinguish what belongs to the idea from what has merely been assumed around it. Gomes et al. (2022) show how physical representations can support understanding and analogical transfer during concept design, while design research more broadly treats representations as cognitive objects that change thinking rather than passive records of finished ideas.

Clarification should therefore remain separate from verbosity and presentation quality. Kornish and Jones (2021) found that more verbose descriptions of raw ideas increased perceived creativity. Their result is methodologically important because a richer description can make an idea feel more developed even when the underlying concept has not improved. Additional words can hide ambiguity as easily as they can resolve it.

Clarification can succeed even when the idea later proves weak, because making its mechanism explicit may be what reveals that it is unworkable. The immediate achievement is intelligibility: the creator becomes able to say what the idea is, what it is trying to change, how its main elements relate and which ambiguities remain unresolved.

The distinction matters because creative work can stall at slogan level. “AI-powered personalised delivery,” “frictionless customer experience” or “community logistics” may sound generative while concealing an absence of mechanism. Creative Capacity has not finished its work when the idea acquires attractive language. It has progressed when the emerging possibility becomes clear enough to examine.

Concept shaping changes the idea itself

Once the idea is intelligible, development increasingly concerns its internal structure. Shaping the concept is the capacity to organise, integrate and revise the elements of a clarified idea into a coherent conceptual form that can be understood, examined, challenged and developed further.

Creativity science offers no universal definition separating an idea from a concept, so we use a functional boundary. A raw idea can be fragmentary, containing a direction, analogy or proposed mechanism. A concept contains a more explicit architecture: central components have been selected, relationships are specified, contradictions have been addressed sufficiently, and the boundaries are stable enough for another person to inspect the structure rather than guess what the creator intended.

Research on refinement and concept development supports this later creative work. Baruah and Green (2023) found that collaborative refinement after earlier generation and selection improved originality, elaboration and quality. Gillier and Bayus (2022) likewise show final concepts emerging through selective integration and development of earlier material, with the earlier ideas changing as they enter the final concept. The creative contribution continues after ideation because changing relationships among components can produce a concept that did not exist in any single starting proposal.

Generation and refinement are also less separable than a strict divergence-convergence model suggests. Baruah, Paulus and Kohn (2021) found benefits when creators attended simultaneously to originality and feasibility during the process, with elaboration contributing to resulting quality. A discovered weakness can trigger another act of generation, while a promising addition can require the concept to be reorganised around it.

Analytical Capacity interacts strongly at this point because detecting a contradiction, testing an assumption or examining whether a relationship is coherent are recognisably analytical operations. Creative shaping remains active when the response alters the representation itself by recombining components, changing a mechanism, simplifying the structure or introducing a new relationship. In practice, interrogation and reconstruction can alternate so quickly that assigning every micro-operation exclusively to one capacity would be artificial.

The logistics team may discover, for example, that small retailers cannot store parcels for long periods. Analytical work reveals a constraint in the emerging model. Creative shaping can respond by changing the architecture from storage to scheduled handoff windows, integrating customer choice with route timing and redefining the role of the local partner. The concept has developed through a change in its internal form, with evaluation serving as information that triggered reconstruction.

Meaning becomes clearer when the concept enters a context

Creativity becomes context dependent as soon as value enters the question. Mainstream definitions of creative products generally combine some form of novelty with usefulness, appropriateness, effectiveness or value, although research shows that these criteria are weighted differently across tasks and domains (Runco and Jaeger, 2012; Harvey and Berry, 2023).

Lloyd-Cox, Pickering and Bhattacharya (2022) provide a useful comparison. Novelty influenced creativity judgments in both Alternate Uses responses and real-world urban-development proposals, while usefulness mattered much more strongly for the real projects. The same criterion therefore acquires different weight once an idea enters a context where people, constraints and consequences are real.

Articulating value and applications is the least directly established of our six dimensions and needs the narrowest interpretation. The defensible function concerns recognising and making explicit plausible ways in which a developed concept could matter, for whom, under which circumstances and towards what kinds of outcomes. It does not include proving that the value exists, persuading others to support the concept, securing resources, implementing it or commercialising it.

Those boundaries prevent a creative dimension from absorbing communication, analysis and execution. A creator can imagine that neighbourhood handoff points might reduce failed deliveries for customers in dense residential areas. That application is plausible at the creative stage, while establishing whether the model actually reduces cost, whether customers will adopt it, whether local merchants will participate and whether regulation permits the arrangement requires evidence and analytical evaluation. Persuading merchants to join or operationalising the network requires additional Practical Skills.

The distinction between recognising value and demonstrating value also protects creativity from retrospective success bias. A plausible application can fail after testing because the creator lacked information that was unavailable during concept development. Conversely, an initially unfashionable use can become valuable when the context changes. Creative Capacity concerns the ability to connect the concept with possible relevance while preserving the difference between a possibility and an established result.

Guo et al.'s (2022) meta-analysis shows why that boundary matters by finding only a small relationship between divergent thinking and evaluative skill, while Puente-Díaz, Cavazos-Arroyo and Puerta-Sierra (2021) show that creators can overestimate the quality of their own ideas and select poorly. Generating a concept and recognising its likely value therefore draw on overlapping but non-identical capabilities. The creator's confidence in an idea is not evidence that its proposed application is valuable.

The word articulating can also mislead if it is read as persuasive communication. In this dimension it refers to making relevance explicit enough to examine. A person can articulate potential value in plain language and still lack the Practical Skill required to pitch the concept effectively. A polished pitch can likewise disguise a concept with little underlying value.

Creative work keeps reopening itself

The six dimensions make the process easier to examine, but a fixed six-stage sequence would misrepresent how creation develops. Contemporary integrative work increasingly treats divergent, convergent, evaluative and elaborative processes as dynamically interwoven rather than allocated to separate chronological blocks (Pinkow, 2023).

Opening the creative space changes what can be generated, but an unexpected idea can reveal that the initial frame should be reconsidered. Generating creates candidates for exploration, while exploring one candidate can create several new ideas. Clarification can expose a hidden assumption that requires reframing. Shaping may produce a new component worth exploring, while thinking about a possible application can reveal that the concept has been built for the wrong context and needs structural revision.

Evaluation operates across this movement from early filtering through later concept development. Some filtering is already present when a person decides that an association is irrelevant or that another variation would simply repeat what has already been produced. Evaluation becomes constraining when it hardens into premature commitment and closes alternative directions before sufficient exploration. It becomes productive when a discovered weakness triggers another round of generation, exploration or shaping rather than merely terminating the idea.

Creative metacognition helps explain why this regulation is difficult. Puente-Díaz, Cavazos-Arroyo and Puerta-Sierra (2021) found that beliefs and judgments about one's own ideas influence later selection, while overestimation can reduce selection accuracy. The person has to regulate transitions among opening, search, development and evaluation while remaining aware that subjective ease, attachment and confidence can mislead those choices.

Different creative episodes can therefore express the six dimensions in very different proportions. A routine design variation may require little reframing and substantial shaping. An innovation problem built around an inherited assumption may depend heavily on Opening the creative space. A concept already supplied by another person can still require exploration, clarification and shaping from someone who did not generate the original idea. The framework describes functions that can be mobilised and revisited, not boxes through which every creator must travel once.

Knowledge gives creativity material and boundaries

A creative process cannot generate meaningful concepts from nothing. Knowledge supplies the materials available for recombination, examples for analogy, causal models for anticipating consequences, standards for recognising novelty and contextual constraints that distinguish an unusual idea from an uninformed one.

Research on domain specificity supports a mixed structure containing cross-domain relationships alongside substantial domain variation. Acar et al. (2024), using meta-analytic confirmatory factor analysis across 30 studies and more than 31,000 participants, found evidence for both general and domain-level structure in self-reported creativity. Its psychometric scope is narrower than our six-dimensional architecture, while its results add substantial evidence that creative expression contains both cross-domain and domain-specific variation. Psychometric critiques similarly warn that conclusions about domain generality depend heavily on how creativity and domain are defined (Myszkowski et al., 2024).

Expertise strengthens high-level creativity because the creator understands what has already been tried, which constraints are real, which combinations are plausible and which apparent innovations are already standard practice. The same organised knowledge can also create fixation by making familiar representations disproportionately accessible. Expertise therefore changes the quality of the search space while creating the possibility that the search stays too local.

A novice in medicine can generate an unusual treatment idea, yet unfamiliarity with physiology, prior evidence and safety constraints prevents the person from knowing whether the idea is coherent, genuinely novel or dangerous. High Creative Capacity cannot substitute for a missing Knowledge Domain. At the same time, domain expertise without the ability to reopen assumptions, generate alternatives or reshape concepts can leave sophisticated knowledge attached to familiar solutions.

Knowledge gives Creative Capacity something substantive to work with, while creative activity can reorganise existing knowledge, expose gaps and generate questions that lead to further learning, creating a reciprocal developmental relationship. The distinction between the two matters because it prevents novelty from being mistaken for expertise and expertise from being treated as creativity.

Context can release creative work or narrow it

Creative Capacity belongs to the individual while its expression remains deeply affected by the environment. Constraints, time pressure, leadership, evaluation, collaboration and psychological safety change which possibilities are explored and which ideas are voiced.

Research on constraints is particularly useful because it rejects the common assumption that creativity increases as rules disappear. Damadzic et al. (2022) show through meta-analysis that constraint effects depend on their form and context, while Tromp (2023) and Cromwell (2024) develop more differentiated accounts of how constraints can structure or suppress creative search. A creator working without meaningful boundaries can face an unusably large possibility space; a creator facing rigid or contradictory restrictions can be trapped inside a space too narrow for exploration.

Keith (2023) found that specifying desired creative outcomes can guide effort differently from prescribing exactly how the creative process should proceed. A useful brief can provide direction while leaving the mechanism open; a highly prescriptive process can reduce autonomy and encourage compliance with the expected method instead of exploration.

Social environments add both cognitive input and social cost. Collaboration can expose creators to associations and domain perspectives they would not retrieve independently. It can also produce production blocking, conformity, evaluation apprehension and status effects. Wang and Peng (2022), for example, found that shared leadership can support creativity through psychological safety while simultaneously creating role stress. Environmental support therefore changes whether Creative Capacity is expressed without becoming a component of the capacity itself.

Emotion alters creative work in context-dependent ways, with positive affect producing no universal increase in creativity and negative affect producing no uniform suppression. Lai et al. (2021) found that affective states influenced ideation and evaluation differently over time. Anxiety can narrow exploration in one setting and increase attention to threat-relevant constraints in another; curiosity can sustain search, frustration can end it, and attachment to an idea can make revision harder. Emotional Capacity contributes by helping the creator remain able to explore, receive feedback and abandon or reshape a concept when emotional responses might otherwise control the process.

Creative Capacity can be developed, although transfer remains the harder test

The intervention literature gives a clear answer to the broad question of malleability: several constituent creative processes can be improved through structured development. The stronger claim that short interventions reliably create durable, domain-general Creative Capacity is much harder to support.

McKay et al.'s (2024) meta-analysis of creativity training in organisational settings is particularly informative. Training produced meaningful overall and learning effects, while the pooled effect on behavioural or on-the-job transfer was smaller and statistically non-significant, and delayed assessments showed substantially weaker effects than immediate measures. Training can therefore improve performance under the conditions it targets without demonstrating that the underlying capacity will transfer broadly or remain unchanged over time.

Gu et al. (2022) found the same process specificity when several brief cognitive creativity-training methods produced different effects on fluency, flexibility and originality, while none improved performance on a convergent Remote Associates outcome. A technique can strengthen a particular search process without lifting every component of Creative Capacity together.

Development is consequently more defensible when it is process specific. Opening can be exercised by identifying inherited assumptions, changing abstraction levels and constructing alternative frames. Generating can be developed through sustained retrieval, category switching, conceptual combination and deliberate variation of mechanisms. Exploring can be practiced by forcing multiple trajectories from one seed idea. Clarifying benefits from externalisation and feedback that expose ambiguity. Shaping requires iterative construction in which components and relationships are challenged and revised. Articulating value and applications can be developed by asking creators to connect concepts with several plausible needs and contexts while separating conjectured relevance from evidence.

The evidence supporting those practices is uneven across dimensions. Generation-related interventions have the most mature research base, while long-term training evidence for Exploring, Clarifying, Shaping and Articulating value is substantially thinner. Authentic projects are likely to matter because later creative functions depend on domain knowledge, constraints and feedback, while evidence of broad transfer requires direct testing beyond successful performance inside the training environment.

What happens after practice also shapes development because feedback can reveal redundancy, incoherence or misplaced confidence, while varied tasks reduce the risk that a technique becomes attached to one familiar prompt. The creator needs opportunities to revise rather than simply receive a score because creative improvement often involves changing the path by which a concept develops.

Assessment has to observe the process as well as the product

Existing creativity measures sample only parts of the capacity we describe. Divergent-thinking tasks are useful for aspects of Generating diverse ideas, but they cannot tell us whether a person can explore an emerging idea, clarify its mechanism, shape a coherent concept or connect it with plausible value.

The scale of that measurement problem is visible in Saretzki, Forthmann and Benedek's (2026) systematic quantitative review of divergent-thinking assessment. They identified 451 eligible studies across 396 articles and more than 110 different administration and scoring conditions. The field has an enormous body of divergent-thinking research without a single stable operationalisation of the process, which makes any equation between a divergent-thinking score and Creative Capacity especially difficult to defend.

Product assessment answers a different question, and the Consensual Assessment Technique and related expert-rating approaches can produce useful judgments about the creativity of outputs, and Long and Wang's (2022) review found substantial evidence of reliability alongside context and validity limitations. Barth and Stadtmann (2021) also show that aggregate consensual ratings can remain highly reliable even when individual raters drift over time. These measures tell us something important about what was produced while revealing much less about how the person produced it.

A credible assessment of Creative Capacity therefore needs repeated tasks and process evidence. Alternative frames and escape from fixation can provide evidence about Opening, while genuinely non-redundant directions help reveal Generating beyond response count. Exploration becomes visible in branches, implications and transformations developed from a seed. Changes in intelligibility, boundaries and mechanism can indicate Clarifying, and concept versions can expose the quality of structural revisions involved in Shaping. Articulating value can be examined through plausible connections to users, needs and contexts without rewarding pitch polish.

Process evidence matters because the same strong product can arise through very different routes. Prior familiarity may supply the solution, another collaborator may contribute the central insight, a template may be adapted, or luck may produce a combination the creator cannot reproduce. A difficult process may also produce a mediocre output when domain knowledge or resources are insufficient. Assessment becomes more defensible when it combines the quality of outcomes with evidence of frames, alternatives, revisions and conceptual development.

AI makes the distinction especially visible because unaided performance, assisted performance and subsequent unaided transfer answer different questions. A person who produces an exceptional AI-assisted concept may demonstrate strong ability to direct, evaluate and shape machine-generated material while providing little evidence about unaided ideation. Another person may use the same tool in a way that exercises reframing and concept development extensively. Recording assistance and provenance is therefore part of understanding what capacity the final product actually evidences.

AI can raise creative performance while changing who did the creative work

Generative AI is unusually useful for studying Creative Capacity because it can improve observable products without telling us whether the human processes underneath them have improved.

Doshi and Hauser (2024) found in a randomised creative-writing study that access to generative-AI ideas increased evaluated story creativity, especially for participants who began with lower creative performance, while the resulting stories became more similar to one another. Lee and Chung (2024), across five experiments, likewise found that ChatGPT could improve rated creativity on several ideation tasks. These studies show real augmentation effects at the level of output.

Other evidence reveals the trade-offs hidden inside average performance. Meincke, Nave and Terwiesch (2025) reanalysed brainstorming data and found lower diversity across the total pool of AI-assisted ideas even where average individual ratings improved. Wadinambiarachchi et al. (2024) found greater fixation and weaker divergent outcomes in a visual ideation setting using AI image generation. Bangerl et al. (2025) report that AI-supported collaboration did not improve every component of creative work and that reduced elaboration in assisted conditions raises questions about what remains when assistance is removed.

These findings are compatible because AI can change different parts of the process in different directions. It can rapidly supply alternatives, analogies and polished formulations, reducing the cost of generation and clarification. Those same outputs can become anchors that narrow independent search. A user can select and reshape machine suggestions actively, or accept coherent-looking material before the underlying concept has been understood. The quality of the final product alone cannot reveal which division of cognitive labour occurred.

The six dimensions make that division easier to inspect. AI can suggest alternative frames while also fixing the user on its first framing. It can generate many candidates while homogenising the pool of possibilities. It can simulate applications while encouraging acceptance of plausible but unsupported claims. It can reorganise a concept while doing much of the synthesis that would otherwise exercise human shaping.

The developmental question remains open because most evidence is recent and short-term. Stronger assisted output does not establish stronger independent capacity, and current research cannot justify confident claims that repeated AI use will either enhance or erode Creative Capacity over the long term. The relevant evidence would need to compare unaided baseline, assisted work, later unaided performance and transfer to genuinely new tasks over meaningful periods.

For now, the safest interpretation is that AI changes the allocation of creative work. Human Creative Capacity becomes especially visible when the person challenges the model's default framing, searches beyond its first suggestions, recognises false coherence, develops fragments into a concept with domain-specific intent and decides which possible applications deserve further investigation.

The creative process continues after novelty appears

The research behind Creative Capacity is strongest around problem construction and idea generation. Evidence for the middle and later dimensions comes from a broader synthesis across elaboration, design cognition, external representation, refinement, evaluation and contextual relevance, leaving the six dimensions with different levels of direct empirical support.

Opening the creative space and Generating diverse ideas have strong direct foundations. Exploring ideas' potential is well supported as a function even though it overlaps with generation, elaboration and refinement. Clarifying the idea rests principally on convergent evidence from externalisation and representational development. Shaping the concept has stronger behavioural support from refinement and design research while remaining intertwined with analysis and domain skill. Articulating value and applications remains defensible when it refers narrowly to identifying plausible relevance while persuasion, validation and implementation remain outside the dimension.

The wider architecture still captures an important part of creative work that conventional idea-generation measures leave behind. A person can produce an original fragment without knowing what to do with it. Meaningful creation requires enough development for the possibility to acquire intelligibility, internal structure and a plausible relationship with a real context. That development depends on knowledge, feedback, constraints, emotion and analytical judgment, while remaining capable of reopening the space and changing the concept itself.

Creative Capacity describes the processes through which a person can keep an emerging possibility open, develop it far enough to understand what it is, reshape it when the concept proves weak and connect it with a context in which its relevance can be examined. The work can begin by challenging a frame, continue through several competing directions and spend substantial time inside one emerging idea before any concept is ready to be judged. Its quality becomes visible in the range of possibilities created and in the creator's ability to develop one of them without confusing novelty with value, polish with clarity or assistance with independent capacity.