The Metacognitive Paradox of AI-Assisted Creativity: A Theoretical Extension
Open accessAbstract
The rapid integration of large language models (LLMs) into organizational workflows raises fundamental questions about the long-term effects of AI assistance on human creative capabilities. This article provides a comprehensive theoretical extension of Sun et al.'s (2025) field experiment, which demonstrated that LLM assistance enhances creative output during use but diminishes subsequent independent creativity—particularly for individuals with lower metacognitive ability. We develop the metacognitive paradox framework to explain this phenomenon: AI tools designed to augment human creativity may inadvertently suppress the very cognitive processes that sustain independent creative capacity. Drawing on metacognition theory, cognitive offloading research, automation and human factors literatures, and the componential model of creativity, we articulate the mechanisms through which LLM assistance affects creative cognition, specify temporal dynamics and boundary conditions, and generate testable propositions for future research. We explicitly compare our framework against alternative theoretical explanations—including cognitive load theory, motivational accounts, and expertise development perspectives—and provide methodological guidance for testing our propositions. Our analysis reveals that the relationship between AI assistance and human creativity is neither uniformly beneficial nor detrimental but contingent upon individual metacognitive capabilities, patterns of tool use, task characteristics, and organizational context. We conclude with implications for theory, research methodology, organizational practice, and the ethical dimensions of AI-augmented work.
AI snapshot
Written from the paper’s full text.
TL;DR
This theoretical article extends Sun et al.’s (2025) field experiment by proposing a metacognitive paradox: LLM assistance can improve immediate creative output while weakening later independent creativity, especially for people with lower metacognitive ability. It specifies mechanisms, moderators, temporal dynamics, and a research agenda for testing the framework.
Key findings
- The paper’s central claim is that AI assistance externalizes metacognitive monitoring and control, yielding short-term performance gains but risking suppression or erosion of independent creative capacity.
- It cites Sun et al. (2025), a field experiment with 656 employees and roughly eight days of observation: LLM-assisted participants performed better on the initial creative task, but later, without LLM access, performed worse than baseline and control; metacognitive ability moderated this pattern, with higher-ability individuals showing smaller decrements, and reduced metacognitive engagement partially mediated the effect.
- The framework proposes that strategic monitoring and metacognitive knowledge mediate effects; assistance intensity may be non-linear; need for cognition, domain expertise, task complexity, assistance type, organizational norms, psychological safety, and creative stage moderate outcomes; suppression is rapid while erosion is gradual; and “metacognitive maintenance intervals” may reduce erosion.
- It compares the framework with cognitive load, motivational, and expertise-development accounts, arguing that metacognitive processes offer a more specific mechanism while other factors may complement it.
Methods
- Conceptual/theoretical extension, not a new empirical study; it synthesizes metacognition theory, cognitive offloading and extended cognition, automation/human factors, and creativity research.
- Uses Sun et al. (2025) as the anchoring field experiment, compares alternative theoretical explanations, and generates testable propositions.
- Offers methodological guidance for future work, including longitudinal field experiments, experience sampling, think-aloud protocols, behavioral trace analysis, randomized withdrawal designs, and construct operationalization.
Limitations
- Stated: the framework is mainly derived from one field experiment; replication is needed across organizational contexts, creative domains, and AI systems; temporal dynamics of erosion and recovery are underspecified; the analysis is individual-level rather than team-level; other individual differences and creative identity remain unexplored; generalizability across different AI systems is uncertain.
- Evident from the text: no new data are collected, the propositions are untested, and the argument depends on secondary interpretation of Sun et al. (2025), so causal claims await empirical verification.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
1 of 35 cited sources checked · 1 supported · 0 flagged
- Interpretationreasoning
AI tools designed to augment human creativity may inadvertently suppress the cognitive processes that sustain independent creative capacity.
“We develop the metacognitive paradox framework to explain this phenomenon: AI tools designed to augment human creativity may inadvertently suppress the very cognitive processes that sustain independent creative capacity.”
Quote in the paperSubstantial claim - Interpretationreasoning
The relationship between AI assistance and human creativity is neither uniformly beneficial nor detrimental but contingent upon individual metacognitive capabilities, patterns of tool use, task characteristics, and organizational context.
“creativity is neither uniformly beneficial nor detrimental but contingent upon individual metacognitive capabilities, patterns of tool use, task characteristics, and organizational context.”
Quote in the paperNot a substantial claim - Premisecitation
Experts possess rich, well-organized knowledge structures that support metacognitive monitoring and control.
“Experts possess rich, well-organized knowledge structures that support metacognitive monitoring and control (Chi et al., 1988).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Chi et al., 1988
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
This cognitive infrastructure may be more resistant to disruption from AI assistance than the less developed knowledge structures of novices.
“This cognitive infrastructure may be more resistant to disruption from AI assistance than the less developed knowledge structures of novices.”
Quote in the paperSubstantial claim - Premisecitation
Mediation analyses in Sun et al. (2025) suggested that reduced metacognitive engagement during LLM-assisted work partially explained the subsequent creativity decrements.
“Mediation analyses suggested that reduced metacognitive engagement during LLM-assisted work partially explained the subsequent creativity decrements.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Interpretationreasoning
Proposition 2b: Need for cognition moderates the effects of LLM assistance on subsequent creative performance, with individuals high in need for cognition experiencing smaller decrements in subsequent creativity following LLM use.
“Proposition 2b: Need for cognition-the tendency to engage in and enjoy effortful cognitive activity (Cacioppo & Petty, 1982)-moderates the effects of LLM assistance on subsequent creative performance.Individuals high in need for cognition experience smaller decrements in subsequent creativity following LLM use.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Cacioppo & Petty, 1982 source
- Interpretationcitation
The pattern that individuals with high metacognitive ability did not show the same decrements in subsequent performance is more consistent with a metacognitive than a purely motivational explanation.
“This pattern is more consistent with a metacognitive than a purely motivational explanation.”
Quote in the paperSubstantial claim- The authors state this as an assumption: Individuals with high metacognitive ability experienced similar motivational dynamics as their lower-metacognitive-ability counterparts.
- Interpretationcitation
AI assistance might undermine expertise development by reducing the deliberate practice that would otherwise occur during creative work.
“From this perspective, AI assistance might undermine expertise development by reducing the deliberate practice that would otherwise occur during creative work.This explanation has merit, particularly for understanding longer-term effects of AI assistance on creative skill development.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Ericsson et al., 1993 source
- Interpretationreasoning
When subsequently required to work without AI assistance, the individual must reconstitute metacognitive processes, often finding that their monitoring is less acute, their strategic knowledge less accessible, and their control less flexible than before AI exposure.
“When subsequently required to work without AI assistance, the individual must reconstitute these metacognitive processes, often finding that their monitoring is less acute, their strategic knowledge less accessible, and their control less flexible than before AI exposure.”
Quote in the paperNot a substantial claim - Premisecitation
Cognitive load theory predicts that removing AI assistance would simply return cognitive load to baseline levels, not produce decrements below baseline performance.
“However, this theory also predicts that removing the AI assistance would simply return cognitive load to baseline levels, not produce decrements below baseline performance.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sweller, 1988 source
- Interpretationreasoning
Proposition 3a: Metacognitive suppression occurs rapidly upon initiation of LLM assistance, while metacognitive erosion develops gradually with sustained or repeated AI use.
“Proposition 3a: Metacognitive suppression (temporary reduction in metacognitive engagement) occurs rapidly upon initiation of LLM assistance, while metacognitive erosion (more persistent changes in metacognitive capability) develops gradually with sustained or repeated AI use.”
Quote in the paperNot a substantial claim - Interpretationreasoning
When LLM assistance becomes available, individuals can offload substantial portions of metacognitive operations to the AI system, reducing immediate cognitive burden while maintaining or enhancing task performance.
“When LLM assistance becomes available, individuals can offload substantial portions of these metacognitive operations to the AI system.Rather than monitoring their own idea generation, they can evaluate AI-generated suggestions.Rather than recognizing impasses, they can query the AI for additional options.Rather than strategically adjusting their approach, they can refine their prompts or accept”
Quote in the paperNot a substantial claim - Interpretationreasoning
Proposition 2c: Domain expertise moderates the relationship between LLM assistance and subsequent creative performance, with experts experiencing smaller decrements than novices because their well-developed domain knowledge provides a stable foundation for metacognitive operations even when AI assistance is removed.
“Proposition 2c: Domain expertise moderates the relationship between LLM assistance and subsequent creative performance.Experts experience smaller decrements than novices because their well-developed domain knowledge provides a stable foundation for metacognitive operations even when AI assistance is removed.”
Quote in the paperSubstantial claim - Premisecitation
Metacognitive skills develop through practice and experience.
“Importantly, metacognitive skills develop through practice and experience (Winne & Nesbit, 2010).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Winne & Nesbit, 2010
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Premisecitation
When individuals relied on AI assistance, they reported less active monitoring and regulation of their own thinking processes.
“When individuals relied on AI assistance, they reported less active monitoring and regulation of their own thinking processes.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Interpretationreasoning
The paradoxical quality of this phenomenon operates at individual, organizational, and societal levels.
“The paradoxical quality of this phenomenon operates at multiple levels.At the individual level, rational behavior (using available tools to enhance performance) produces seemingly irrational outcomes (diminished capability over time).At the organizational level, investments in AI tools to boost creativity may paradoxically undermine the human creative capital that organizations depend upon.At the”
Quote in the paperSubstantial claim - Interpretationreasoning
Reduced strategic monitoring during LLM-assisted work leads to diminished monitoring capabilities in subsequent unassisted work.
“Specifically, reduced strategic monitoring during LLM-assisted work leads to diminished monitoring capabilities in subsequent unassisted work.”
Quote in the paperNot a substantial claim - Interpretationreasoning
Proposition 1b: Metacognitive knowledge is less actively accessed and refined during LLM-assisted creative work, resulting in less available and less flexible strategic knowledge during subsequent unassisted work.
“Proposition 1b: Metacognitive knowledge-understanding of effective creative strategies and when to deploy them-is less actively accessed and refined during LLM-assisted creative work, resulting in less available and less flexible strategic knowledge during subsequent unassisted work.”
Quote in the paperNot a substantial claim - Interpretationreasoning
The metacognitive paradox framework accounts for rapid effects by focusing on the immediate disengagement of metacognitive processes rather than the longer-term accumulation of expertise.
“The metacognitive paradox framework accounts for these rapid effects by focusing on the immediate disengagement of metacognitive processes rather than the longer-term accumulation of expertise.”
Quote in the paperSubstantial claim - Premisecitation
In Sun et al. (2025), when subsequently asked to complete a creative task without LLM access, those who had previously used AI assistance showed diminished creative performance relative to their baseline capabilities and to the control group's subsequent performance.
“However, when subsequently asked to complete a creative task without LLM access, those who had previously used AI assistance showed diminished creative performance relative to their baseline capabilities and to the control group's subsequent performance.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Premisecitation
Reduced metacognitive engagement appeared to have carry-over effects, diminishing subsequent independent creative performance.
“This reduced metacognitive engagement appeared to have carry-over effects, diminishing subsequent independent creative performance.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Interpretationcitation
The Sun et al. (2025) findings, observed over approximately eight days, likely reflect a combination of metacognitive suppression and metacognitive erosion.
“The Sun et al. (2025) findings, observed over approximately eight days, likely reflect a combination of these processes.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Premisecitation
In Sun et al. (2025), individuals assigned to LLM assistance demonstrated enhanced creative performance on the initial task compared to those working without AI support.
“Individuals assigned to use LLM assistance demonstrated enhanced creative performance on the initial task compared to those working without AI support.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Interpretationreasoning
Proposition 1c: The relationship between LLM assistance intensity and subsequent creative performance follows a non-linear pattern, with moderate assistance potentially enhancing subsequent performance while intensive assistance diminishes it.
“The relationship between LLM assistance intensity and subsequent creative performance follows a non-linear pattern, with moderate assistance potentially enhancing subsequent performance while intensive assistance diminishes it.”
Quote in the paperNot a substantial claim - Interpretationcitation
Proposition 2a: Individuals with higher baseline metacognitive ability maintain more stable metacognitive engagement during LLM-assisted work and demonstrate greater resilience in subsequent unassisted creative performance.
“Proposition 2a: Individuals with higher baseline metacognitive ability maintain more stable metacognitive engagement during LLM-assisted work and demonstrate greater resilience in subsequent unassisted creative performance.”
Quote in the paperSubstantial claim - Interpretationreasoning
Proposition 1a: Strategic monitoring mediates the relationship between LLM assistance and subsequent creative performance.
“Proposition 1a: Strategic monitoring-the ongoing assessment of progress toward creative goalsmediates the relationship between LLM assistance and subsequent creative performance.”
Quote in the paperSubstantial claim - Interpretationnone given
The core mechanism underlying the metacognitive paradox involves the externalization of metacognitive functions to AI systems.
“The core mechanism underlying the metacognitive paradox involves the externalization of metacognitive functions to AI systems.”
Quote in the paperNot a substantial claim - Interpretationreasoning
The effects observed in Sun et al.'s (2025) study may combine rapid suppression with incipient erosion.
“This proposition suggests that the effects observed in Sun et al.'s (2025) study may combine rapid suppression with incipient erosion.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- Interpretationreasoning
Organizations that frame AI as a complement to, rather than substitute for, human cognition may foster more reflective engagement that preserves metacognitive functioning.
“Organizations that frame AI as a complement to, rather than substitute for, human cognition may foster more reflective engagement that preserves metacognitive functioning.”
Quote in the paperNot a substantial claim- Rests on an assumption: Organizational context shapes how individuals approach AI tools.
- Interpretationreasoning
The effects of LLM assistance on metacognitive engagement and subsequent creative performance differ across stages of the creative process.
“Proposition 5: The effects of LLM assistance on metacognitive engagement and subsequent creative performance differ across stages of the creative process.”
Quote in the paperSubstantial claim - Interpretationreasoning
The distribution of cognition across person and technology has implications for what happens when that distribution changes.
“However, we emphasize that the distribution of cognition across person and technology has implications for what happens when that distribution changes.”
Quote in the paperNot a substantial claim - Interpretationreasoning
In psychologically safe environments, individuals are more willing to acknowledge uncertainty, seek feedback, and maintain metacognitive engagement even when AI assistance is available.
“In psychologically safe environments, individuals are more willing to acknowledge uncertainty, seek feedback, and maintain metacognitive engagement even when AI assistance is available.”
Quote in the paperNot a substantial claim - Interpretationreasoning
When individuals feel safe to admit what they do not know and to seek help in developing their own capabilities, they may be more resistant to the complacency that can accompany AI assistance.
“When individuals feel safe to admit what they do not know and to seek help in developing their own capabilities, they may be more resistant to the complacency that can accompany AI assistance.”
Quote in the paperNot a substantial claim - Interpretationreasoning
Generative assistance may allow more complete externalization of metacognitive processes than evaluative assistance, which still requires the individual to produce content for evaluation.
“Generative assistance may allow more complete externalization of metacognitive processes than evaluative assistance, which still requires the individual to produce content for evaluation.”
Quote in the paperNot a substantial claim - Interpretationreasoning
If metacognitive processes become distributed across person and AI system, removing the AI system does not simply return the individual to their prior state.
“If metacognitive processes become distributed across person and AI system, removing the AI system does not simply return the individual to their prior state.”
Quote in the paperSubstantial claim - Interpretationreasoning
LLM assistance that displaces generative processes may have different consequences than assistance that supports evaluative processes.
“LLM assistance that displaces generative processes may have different consequences than assistance that supports evaluative processes.”
Quote in the paperSubstantial claim - Interpretationreasoning
The extended mind, once contracted, may not readily re-expand to its former boundaries.
“The extended mind, once contracted, may not readily re-expand to its former boundaries.”
Quote in the paperNot a substantial claim - Premisecitation
Experimental manipulation of the mediator provides stronger causal evidence than correlational mediation analysis.
“Experimental manipulation of the mediator provides stronger causal evidence than correlational mediation analysis (Spencer et al., 2005)”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Spencer et al., 2005 source
- Interpretationreasoning
For complex creative tasks requiring substantial metacognitive engagement, the effects of AI assistance on subsequent performance are larger than for simpler tasks.
“For complex creative tasks requiring substantial metacognitive engagement, the effects of AI assistance on subsequent performance are larger than for simpler tasks.”
Quote in the paperNot a substantial claim- Rests on an assumption: Complex creative tasks require more metacognitive resources than simple tasks.
- Interpretationcitation
If AI assistance is differentially beneficial for high-skill individuals (as Sun et al.'s findings suggest), widespread AI deployment may exacerbate rather than reduce skill-based inequalities.
“If AI assistance is differentially beneficial for high-skill individuals (as Sun et al.'s findings suggest), widespread AI deployment may exacerbate rather than reduce skill-based inequalities.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al.'s findings
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
Task complexity moderates the relationship between LLM assistance and subsequent creative performance.
“Proposition 4a: Task complexity moderates the relationship between LLM assistance and subsequent creative performance.”
Quote in the paperNot a substantial claim- Rests on an assumption: Complex creative tasks require more metacognitive resources than simple tasks.
- Interpretationreasoning
Organizational norms regarding AI use moderate individual-level metacognitive dynamics.
“Proposition 4c: Organizational norms regarding AI use moderate individual-level metacognitive dynamics.”
Quote in the paperNot a substantial claim- Rests on an assumption: Organizational context shapes how individuals approach AI tools.
- Interpretationreasoning
In organizations that emphasize critical engagement with AI outputs and maintain expectations for independent creative capability, individuals experience smaller decrements in subsequent creativity following LLM use.
“In organizations that emphasize critical engagement with AI outputs and maintain expectations for independent creative capability, individuals experience smaller decrements in subsequent creativity following LLM use.”
Quote in the paperNot a substantial claim- Rests on an assumption: Organizational context shapes how individuals approach AI tools.
- Interpretationreasoning
Proposition 3b: Recovery of metacognitive engagement following LLM use depends on the duration and intensity of prior AI assistance.
“Proposition 3b: Recovery of metacognitive engagement following LLM use depends on the duration and intensity of prior AI assistance.”
Quote in the paperNot a substantial claim- Rests on an assumption: The dose-response relationship between AI assistance and metacognitive effects remains poorly specified.
- Interpretationreasoning
Psychological safety moderates the relationship between LLM assistance and subsequent creative performance.
“Proposition 4d: Psychological safety (Edmondson, 1999) moderates the relationship between LLM assistance and subsequent creative performance.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Edmondson, 1999 source
- Classified as probably not a substantial claim. 59
- May combine two claims. 35
- The quote only partly states this claim. 2
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationHuman-AI teaming frameworks conceptualize the relationship between humans and AI systems as collaborative rather than merely assistive.
“Emerging frameworks on human-AI teaming (Seeber et al., 2020) conceptualize the relationship between humans and AI systems as collaborative rather than merely assistive.”
Quote in the paperNot a substantial claim1 of 1 sources checked- cites Seeber et al., 2020 source
May combine two claims.
PremisecitationLLM assistance enhances creative output during use but diminishes subsequent independent creativity, particularly for individuals with lower metacognitive ability.
“Sun et al.'s (2025) field experiment, which demonstrated that LLM assistance enhances creative output during use but diminishes subsequent independent creativity-particularly for individuals with lower metacognitive ability.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
Classified as probably not a substantial claim. May combine two claims.
PremisecitationIn creative contexts, metacognition manifests as awareness of one's creative strengths and limitations, monitoring of idea quality during generation, and strategic regulation of creative approaches based on task demands and feedback.
“In creative contexts, metacognition manifests as awareness of one's creative strengths and limitations, monitoring of idea quality during generation, and strategic regulation of creative approaches based on task demands and feedback (Davidson & Sternberg, 1998).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Davidson & Sternberg, 1998 source
Probably not a substantial claim: may not pass S2 (checkable).
PremisecitationMetacognition encompasses the knowledge, monitoring, and regulation of one's own cognitive processes.
“Metacognition-cognition about cognition-encompasses the knowledge, monitoring, and regulation of one's own cognitive processes (Flavell, 1979).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Flavell, 1979 source
Classified as probably not a substantial claim.
InterpretationreasoningIndividuals high in need for cognition may find purely receptive interaction with AI unsatisfying and actively seek opportunities for mental effort.
“These individuals may find purely receptive interaction with AI unsatisfying and actively seek opportunities for mental effort.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationEndsley and Kiris (1995) documented the out-of-the-loop performance problem, whereby operators who rely on automation lose situational awareness and struggle to resume manual control when needed.
“Endsley and Kiris (1995) documented the”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Endsley and Kiris (1995) source
Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.
InterpretationreasoningIf the observed effects primarily reflect suppression, brief periods of unassisted work may be sufficient to restore metacognitive engagement; if they reflect erosion, more intensive remediation including explicit metacognitive training may be necessary.
“If the observed effects primarily reflect suppression, brief periods of unassisted work may be sufficient to restore metacognitive engagement.If they reflect erosion, more intensive remediation-including explicit metacognitive training-may be necessary.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationParasuraman and Riley (1997) identified patterns of automation use, misuse, disuse, and abuse.
“Parasuraman and Riley (1997) identified patterns of automation use, misuse, disuse, and abuse that illuminate the complex dynamics of human-technology interaction.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Parasuraman and Riley (1997) source
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationSun (2024) advanced a componential and functional framework for metacognition that identifies three core elements: metacognitive knowledge, metacognitive monitoring, and metacognitive control.
“Sun (2024) advanced a componential and functional framework for metacognition that identifies three core elements: metacognitive knowledge (understanding of cognitive processes and strategies), metacognitive monitoring (ongoing assessment of cognitive states and progress), and metacognitive control (strategic regulation based on monitoring outcomes).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Sun (2024) source
Classified as probably not a substantial claim.
PremisecitationIn Sun et al. (2025), participants completed creative writing tasks evaluated by supervisors and external raters blind to condition.
“Participants completed creative writing tasks that were evaluated by both supervisors and external raters blind to condition.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
InterpretationreasoningThe metacognitive paradox differs from simple skill atrophy or disuse because the metacognitive processes that enable flexible, adaptive creativity become externalized to the AI system, and when the AI is removed, individuals must reconstitute these processes, a task for which they may be less prepared than if they had maintained continuous metacognitive engagement.
“This paradox differs from simple skill atrophy or disuse.The mechanism is not merely that creative skills deteriorate from lack of practice.Rather, the metacognitive processes that enable flexible, adaptive creativity-processes of monitoring, evaluation, and strategic adjustment-become externalized to the AI system.When the AI is removed, individuals must reconstitute these processes, a task for”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningMetacognitive processes represent the primary mechanism, while cognitive load, motivational, and expertise development factors operate as complementary influences that may amplify or attenuate the core metacognitive dynamics.
“We do not claim that the metacognitive paradox framework entirely supersedes these alternative explanations.Rather, we propose that metacognitive processes represent the primary mechanism, while cognitive load, motivational, and expertise development factors operate as complementary influences that may amplify or attenuate the core metacognitive dynamics.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationcitationThe Sun et al. (2025) study observed effects over a relatively brief period (approximately eight days), suggesting that mechanisms beyond gradual skill development are involved.
“However, the Sun et al. (2025) study observed effects over a relatively brief period (approximately eight days), suggesting that mechanisms beyond gradual skill development are involved.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
Probably not a substantial claim: may not pass S4 (based). May combine two claims.
InterpretationreasoningApplied to creative work, automation complacency might manifest as reduced critical evaluation of AI-generated outputs and diminished engagement with the creative process itself.
“Applied to creative work, automation complacency might manifest as reduced critical evaluation of AIgenerated outputs and diminished engagement with the creative process itself.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningThe metacognitive paradox is the phenomenon whereby AI tools designed to enhance human creativity may inadvertently suppress the cognitive processes that sustain independent creative capacity.
“We introduce the term metacognitive paradox to describe the phenomenon whereby AI tools designed to enhance human creativity may inadvertently suppress the cognitive processes that sustain independent creative capacity.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim. The quote only partly states this claim.
PremisecitationSun et al. (2025) measured metacognitive ability using an adapted scale assessing participants' tendencies to plan, monitor, and evaluate their cognitive processes during complex tasks.
“The authors measured metacognitive ability using an adapted scale assessing participants' tendencies to plan, monitor, and evaluate their cognitive processes during complex tasks.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
May combine two claims. The quote only partly states this claim.
InterpretationreasoningThe metacognitive paradox framework offers advantages over alternative explanations in specificity and mechanism precision.
“The metacognitive paradox framework offers several advantages over these alternative explanations:”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationcitationThe Sun et al. (2025) findings specifically implicate metacognitive ability as a moderator, suggesting that cognitive rather than purely motivational processes are centrally involved.
“However, the Sun et al. (2025) findings specifically implicate metacognitive ability as a moderator, suggesting that cognitive rather than purely motivational processes are centrally involved.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
- The authors state this as an assumption: Individuals with high metacognitive ability experienced similar motivational dynamics as their lower-metacognitive-ability counterparts.
May combine two claims.
PremisecitationSkills can atrophy relatively quickly when not regularly exercised, and regaining proficiency after periods of automation-supported performance requires substantial retraining.
“This research, primarily conducted in aviation and industrial control contexts, demonstrates that skills can atrophy relatively quickly when not regularly exercised, and that regaining proficiency after periods of automation-supported performance requires substantial retraining.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Endsley and Kiris (1995) source
May combine two claims.
PremisecitationCognitive offloading, while beneficial for immediate performance, may carry costs for long-term cognitive development; routine reliance on external resources may reduce development of internal capabilities.
“Recent research suggests that cognitive offloading, while beneficial for immediate performance, may carry costs for long-term cognitive development.When individuals routinely rely on external resources for cognitive tasks, they may experience reduced development of internal capabilities (Risko & Gilbert, 2016).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Risko & Gilbert, 2016 source
May combine two claims.
InterpretationreasoningThe metacognitive paradox framework provides a more specific mechanism than cognitive load theory: it identifies the particular cognitive processes (metacognitive monitoring and control) that are disrupted and explains why this disruption persists beyond the period of AI assistance.
“The metacognitive paradox framework provides a more specific mechanism than cognitive load theory: it identifies the particular cognitive processes (metacognitive monitoring and control) that are disrupted and explains why this disruption persists beyond the period of AI assistance.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningThe externalization of metacognitive processes to AI comes at a cost: the metacognitive processes that would normally be exercised and refined during creative work are bypassed, and over time the individual's metacognitive engagement becomes attenuated.
“However, this externalization comes at a cost.The metacognitive processes that would normally be exercised and refined during creative work are instead bypassed.Over time-potentially even over relatively short periods of consistent AI use-the individual's metacognitive engagement becomes attenuated.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningThe metacognitive implications of AI assistance likely differ across automation levels, with higher automation potentially producing greater disengagement and more rapid skill degradation.
“The metacognitive implications likely differ across these levels, with higher automation potentially producing greater disengagement and more rapid skill degradation.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationnone givenThe optimal level of assistance likely varies across individuals and contexts.
“The optimal level of assistance likely varies across individuals and contexts, a boundary condition we elaborate below.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
PremisecitationAutomation complacency describes the tendency for operators to reduce their monitoring and vigilance when automated systems are functioning.
“The concept of automation complacency (Parasuraman & Manzey, 2010) describes the tendency for operators to reduce their monitoring and vigilance when automated systems are functioning.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Parasuraman & Manzey, 2010 source
Probably not a substantial claim: may not pass S4 (based).
InterpretationreasoningHigh need for cognition may lead individuals to maintain cognitive engagement even when external assistance is available, thereby preserving metacognitive functioning.
“High need for cognition may lead individuals to maintain cognitive engagement even when external assistance is available, thereby preserving metacognitive functioning.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningThe metacognitive paradox emerges from a fundamental tension: the efficiency gains that make AI assistance attractive simultaneously reduce opportunities for the metacognitive engagement that maintains and develops creative skills.
“The paradox emerges from a fundamental tension: the efficiency gains that make AI assistance attractive simultaneously reduce opportunities for the metacognitive engagement that maintains and develops creative skills.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationSun et al. (2025) found that metacognitive ability moderated the effects: individuals with higher metacognitive ability maintained stronger subsequent creative performance after LLM use, while those with lower metacognitive ability experienced more pronounced decrements.
“Critically, Sun et al. (2025) found that metacognitive ability moderated these effects.Individuals with higher metacognitive ability maintained stronger subsequent creative performance after LLM use, while those with lower metacognitive ability experienced more pronounced decrements.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
May combine two claims.
InterpretationreasoningMetacognitive suppression may occur rapidly and recover quickly once AI assistance is removed, while metacognitive erosion involves more gradual changes that require more extensive intervention to reverse.
“Suppression may occur rapidly and recover quickly once AI assistance is removed, while erosion involves more gradual changes that require more extensive intervention to reverse.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
PremisecitationThe extended mind thesis proposes that cognitive processes can genuinely extend beyond the boundaries of the individual brain to incorporate external tools and resources.
“The extended mind thesis (Clark & Chalmers, 1998) proposes that cognitive processes can genuinely extend beyond the boundaries of the individual brain to incorporate external tools and resources.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Clark & Chalmers, 1998 source
Probably not a substantial claim: may not pass S3 (load-bearing).
Interpretationnone givenMotivational mechanisms likely contribute to the observed effects of LLM assistance on subsequent creativity.
“We acknowledge that motivational mechanisms likely contribute to the observed effects.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationSun et al.'s (2025) study was conducted with 656 employees of an online education company and employed random assignment to LLM assistance versus control conditions with a design enabling assessment of both concurrent and subsequent effects on creative performance.
“Sun et al.'s (2025) study represents a significant methodological advance in understanding AI effects on creativity.Conducted with 656 employees of an online education company, the experiment employed random assignment to LLM assistance versus control conditions with a design that enabled assessment of both concurrent and subsequent effects on creative performance.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al. (2025) source
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Interpretationnone givenMeasuring metacognitive processes may alter those processes.
“Measurement reactivity: Measuring metacognitive processes may alter those processes.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationcitationEarly-stage ideation may be particularly dependent on generative metacognitive processes-exploration, incubation, insight-while later stages involve more evaluative and refinement processes.
“Early-stage ideation may be particularly dependent on generative metacognitive processes-exploration, incubation, insight-while later stages involve more evaluative and refinement processes.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim. May combine two claims.
InterpretationreasoningThe framework specifies the conditions under which assistance enhances or diminishes human creativity and articulates the cognitive processes through which these effects occur.
“Rather than treating AI effects as uniformly positive or negative, the framework specifies the conditions under which assistance enhances or diminishes human creativity and articulates the cognitive processes through which these effects occur.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone givenThe analysis focuses on individual-level processes.
“Third, our analysis focuses on individual-level processes.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
InterpretationreasoningDistinguishing metacognitive suppression from erosion has important implications for understanding recovery trajectories and designing interventions.
“Distinguishing these processes has important implications for understanding recovery trajectories and designing interventions.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningThe findings suggest value in AI systems designed to support rather than supplant human metacognition.
“The findings suggest value in AI systems designed to support rather than supplant human metacognition.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Interpretationnone givenThe temporal dynamics of metacognitive erosion and recovery remain underspecified.
“Second, the temporal dynamics of metacognitive erosion and recovery remain underspecified.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
InterpretationreasoningIntentional metacognitive engagement, strategic patterns of AI use, organizational support for independent creative work, and AI systems designed to preserve human metacognition can all help realize the benefits of AI assistance while mitigating its costs.
“Intentional metacognitive engagement, strategic patterns of AI use, organizational support for independent creative work, and AI systems designed to preserve human metacognition can all help realize the benefits of AI assistance while mitigating its costs.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
PremisecitationPerry-Smith and Mannucci's (2017) model of the creative process identifies distinct stages with different requirements.
“Perry-Smith and Mannucci's (2017) model of the creative process identifies distinct stages with different requirements.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Perry-Smith and Mannucci's (2017) source
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.
Interpretationnone givenSustained AI use may affect not only cognitive capabilities but also individuals' self-concepts as creative professionals, with potential consequences for motivation and career development.
“Sustained AI use may affect not only cognitive capabilities but also individuals' self-concepts as creative professionals, with potential consequences for motivation and career development.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningOrganizations can mitigate metacognitive erosion through structured alternation between assisted and unassisted work.
“This proposition has direct practical implications, suggesting that organizations can mitigate metacognitive erosion through structured alternation between assisted and unassisted work.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
Interpretationnone givenThe externalization of metacognitive functions to AI systems represents a qualitatively different phenomenon from simple cognitive offloading, with potentially more consequential implications for long-term cognitive development.
“The externalization of metacognitive functions to AI systems represents a qualitatively different phenomenon from simple cognitive offloading, with potentially more consequential implications for long-term cognitive development.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Study design and conductnone givenThe framework is primarily derived from a single, albeit rigorous, field experiment.
“First, our framework is primarily derived from a single, albeit rigorous, field experiment.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
InterpretationreasoningSimilar dynamics to those in procedural tasks in high-reliability contexts operate in knowledge work, with potentially broader implications given the centrality of creativity to organizational adaptation and competitiveness.
“While much of this research has focused on procedural tasks in high-reliability contexts (aviation, industrial control), the present analysis suggests that similar dynamics operate in knowledge work-with potentially broader implications given the centrality of creativity to organizational adaptation and competitiveness.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Interpretationnone givenThe framework distinguishes between suppression and erosion, providing a framework for understanding both rapid and gradual effects.
“Temporal dynamics: It distinguishes between suppression and erosion, providing a framework forunderstanding both rapid and gradual effects.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningSystems that prompt users to evaluate outputs, explain reasoning, and reflect on alternatives may preserve metacognitive engagement better than systems optimized purely for output quality.
“Systems that prompt users to evaluate outputs, explain reasoning, and reflect on alternatives may preserve metacognitive engagement better than systems optimized purely for output quality.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningUnderstanding AI's effects on human cognition requires moving beyond simple questions of whether AI helps or hurts performance to examine the specific cognitive processes through which AI affects human capability.
“The present analysis suggests that this understanding requires moving beyond simple questions of whether AI helps or hurts performance to examine the specific cognitive processes through which AI affects human capability.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
InterpretationreasoningPsychological safety may encourage the kind of reflective practice that sustains metacognitive capability.
“Psychological safety may encourage the kind of reflective practice that sustains metacognitive capability.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
InterpretationreasoningBoth duration and intensity of AI assistance matter, with their combined effect determining the difficulty of recovery.
“This proposition suggests that both duration and intensity matter, with their combined effect determining the difficulty of recovery.”
Quote in the paperNot a substantial claim- Rests on an assumption: The dose-response relationship between AI assistance and metacognitive effects remains poorly specified.
Classified as probably not a substantial claim.
PremisecitationCognitive processes can genuinely extend to incorporate external tools and resources.
“We do not dispute that cognitive processes can genuinely extend to incorporate external tools and resources.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Clark & Chalmers, 1998 source
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationnone givenThe framework connects to the broader literature on automation and skill degradation, providing a more comprehensive account of human-AI interaction.
“Integration with AI research: It connects to the broader literature on automation and skill degradation, providing a more comprehensive account of human-AI interaction.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningCognitive extension is not without costs, and those costs become apparent when extension is disrupted.
“This observation suggests a refinement of the extended mind thesis: cognitive extension is not without costs, and those costs become apparent when extension is disrupted.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningAssistance during early-stage ideation produces different effects than assistance during later-stage refinement and implementation, with early-stage assistance potentially having larger effects on the development of generative metacognitive skills.
“Assistance during early-stage ideation produces different effects than assistance during later-stage refinement and implementation, with early-stage assistance potentially having larger effects on the development of generative metacognitive skills.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningEvaluating AI effects on creativity solely in terms of immediate output quality may miss important consequences for the sustainability of human creative capability.
“Evaluating AI effects on creativity solely in terms of immediate output quality may miss important consequences for the sustainability of human creative capability.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningThe metacognitive paradox framework advances understanding of human-AI interaction in creative work by providing a specific, mechanistic account of how AI assistance affects human cognitive capabilities.
“The metacognitive paradox framework advances understanding of human-AI interaction in creative work by providing a specific, mechanistic account of how AI assistance affects human cognitive capabilities.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim. May combine two claims.
Interpretationnone givenTesting the metacognitive mechanism requires establishing that (a) AI assistance affects metacognitive engagement, (b) metacognitive engagement affects subsequent creativity, and (c) controlling for metacognitive engagement reduces the AI assistance effect.
“Mediation analysis limitations: Testing the metacognitive mechanism requires establishing that (a) AI assistance affects metacognitive engagement, (b) metacognitive engagement affects subsequent creativity, and (c) controlling for metacognitive engagement reduces the AI assistance effect.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Interpretationnone givenWidespread dependence on AI assistance could create workforce vulnerabilities-pools of workers who cannot function effectively without AI support.
“Widespread dependence on AI assistance could create workforce vulnerabilities-pools of workers who cannot function effectively without AI support.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationSun et al.'s (2025) field experiment provides crucial evidence that AI assistance enhances immediate performance while potentially undermining the cognitive foundations of independent creativity.
“Sun et al.'s (2025) field experiment provides crucial evidence that these implications are more complex than early enthusiasm suggested: AI assistance enhances immediate performance while potentially undermining the cognitive foundations of independent creativity.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Sun et al.'s (2025) source
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
InterpretationreasoningOrganizations that mandate or encourage AI use may bear some responsibility for resulting skill effects.
“Organizations that mandate or encourage AI use may bear some responsibility for resulting skill effects.”
Quote in the paperNot a substantial claim- Rests on an assumption: Individuals may have limited awareness of their own capability changes.
Probably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningThe framework contributes to the growing literature on automation and human factors by extending concepts like automation complacency and skill degradation to cognitive and creative work contexts.
“The framework also contributes to the growing literature on automation and human factors by extending concepts like automation complacency and skill degradation to cognitive and creative work contexts.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningThe framework contributes to metacognition theory by identifying a new context-AI-assisted work-in which metacognitive processes operate and can be disrupted.
“The framework contributes to metacognition theory by identifying a new context-AI-assisted workin which metacognitive processes operate and can be disrupted.”
Quote in the paperNot a substantial claim- Rests on an assumption: Prior research has examined metacognition in educational and problem-solving contexts, but AI assistance creates novel dynamics that existing theories have not fully addressed.
Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.
InterpretationreasoningBy identifying the externalization of metacognitive functions to AI systems as the core mechanism, the framework explains both why AI assistance produces immediate benefits and why those benefits may come at a cost to sustained creative capability.
“By identifying the externalization of metacognitive functions to AI systems as the core mechanism, the framework explains both why AI assistance produces immediate benefits and why those benefits may come at a cost to sustained creative capability.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningFor sustainable cognitive augmentation, the human component of the extended system must maintain sufficient capability to function independently when technology becomes unavailable-whether due to technical failure, policy change, or shifting task demands.
“For sustainable cognitive augmentation, the human component of the extended system must maintain sufficient capability to function independently when technology becomes unavailable-whether due to technical failure, policy change, or shifting task demands.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenChanges in metacognitive functioning over time may reflect maturational processes, practice effects, or historical events rather than AI assistance effects.
“Temporal confounds: Changes in metacognitive functioning over time may reflect maturational processes, practice effects, or historical events rather than AI assistance effects.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Interpretationnone givenComplex, iterative queries may indicate more active metacognitive processing than simple, one-shot prompts.
“Complex, iterative queries may indicate more active metacognitive processing than simple, one-shot prompts.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
InterpretationreasoningThis dependence raises concerns about technological lock-in and reduced organizational resilience.
“This dependence raises concerns about technological lock-in and reduced organizational resilience.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
InterpretationreasoningIndividuals with lower baseline metacognitive ability may benefit from additional scaffolding, monitoring, and training to maintain metacognitive engagement during AI-assisted work.
“Individuals with lower baseline metacognitive ability may benefit from additional scaffolding, monitoring, and training to maintain metacognitive engagement during AI-assisted work.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
InterpretationreasoningDifferences among AI systems may affect the metacognitive dynamics described.
“These differences may affect the metacognitive dynamics we describe.”
Quote in the paperNot a substantial claim- Rests on an assumption: ChatGPT, Claude, and specialized creative AI tools differ in their capabilities, interfaces, and interaction patterns.
May combine two claims.
InterpretationreasoningThe individual must reconstitute processes that have been externalized, and they may find their capacity for this reconstitution diminished.
“Rather, the individual must reconstitute processes that have been externalized, and they may find their capacity for this reconstitution diminished.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenEven with random assignment to AI assistance conditions, individuals may self-select into different patterns of AI use.
“Selection effects: Even with random assignment to AI assistance conditions, individuals may self-select into different patterns of AI use.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing), S4 (based). May combine two claims.
Interpretationnone givenCognitive styles, personality factors, prior technology experience, and domain expertise may all moderate the effects of AI assistance on metacognition and creativity.
“Cognitive styles, personality factors, prior technology experience, and domain expertise may all moderate the effects of AI assistance on metacognition and creativity.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningThe metacognitive paradox framework developed in this article offers a theoretical lens for understanding these dynamics.
“The metacognitive paradox framework developed in this article offers a theoretical lens for understanding these dynamics.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
Interpretationnone givenThe framework accounts for why metacognitive ability moderates the effects, a pattern less easily explained by cognitive load or motivational theories.
“It accounts for why metacognitive ability moderates the effects, a pattern less easily explained by cognitive load or motivational theories.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenThe moderating role of metacognitive ability suggests that these costs fall disproportionately on those least equipped to bear them.
“The moderating role of metacognitive ability suggests that these costs fall disproportionately on those least equipped to bear them.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
InterpretationreasoningThe dynamics observed in creative writing may differ from those in visual design, scientific problem-solving, or product development.
“The dynamics observed in creative writing may differ from those in visual design, scientific problem-solving, or product development.”
Quote in the paperNot a substantial claim
The authors state this as an assumption.
Individuals with high metacognitive ability experienced similar motivational dynamics as their lower-metacognitive-ability counterparts.
“Individuals with high metacognitive ability, who presumably experienced similar motivational dynamics as their lower-metacognitive-ability counterparts, did not show the same decrements in subsequent performance.”
The paper relies on this without support.
The dose-response relationship between AI assistance and metacognitive effects remains poorly specified.
“The dose-response relationship between AI assistance and metacognitive effects remains poorly specified.”
The paper relies on this without support.
Complex creative tasks require more metacognitive resources than simple tasks.
“Complex tasks typically demand more metacognitive resources-more monitoring, more strategic adjustment, more recognition of impasses.”
The paper relies on this without support.
Organizational context shapes how individuals approach AI tools.
“Organizational context shapes how individuals approach AI tools.”
The paper relies on this without support.
Prior research has examined metacognition in educational and problem-solving contexts, but AI assistance creates novel dynamics that existing theories have not fully addressed.
“While prior research has examined metacognition in educational and problem-solving contexts, the introduction of AI assistance creates novel dynamics that existing theories have not fully addressed.”
The paper relies on this without support.
Individuals may have limited awareness of their own capability changes.
“Individuals may have limited awareness of their own capability changes.”
The paper relies on this without support.
ChatGPT, Claude, and specialized creative AI tools differ in their capabilities, interfaces, and interaction patterns.
“ChatGPT, Claude, and specialized creative AI tools differ in their capabilities, interfaces, and interaction patterns.”
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 1
- [1]1. Amabile, T. M. (1988). A model of creativity and innovation in organizations. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 10, pp. 123-167). JAI Press.
- Authors
- Westover, J. H.
- Corresponding author
- Westover, Jonathan H.
- Venue
- Innovation: The Journal of Knowledge Work, Creativity, and Organizational Advancement
- Publisher
- Human Capital Innovations, LLC
- Published
- 2026-08
- Volume
- 1
- Issue
- 1
- ISSN
- 3143-8210 (ISSN-L)
- OpenAlex
- W7172312574
- Cited by
- 0
- References
- 0
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Innovation The Journal of Knowledge Work Creativity and Organizational Advancement (publishedVersion)
- SSRN Electronic Journal (acceptedVersion)
- Topics
- Social Sciences › Psychology › Experimental and Cognitive Psychology › Creativity in Education and Neuroscience
- Physical Sciences › Engineering › Mechanical Engineering › Design Education and Practice
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- AI-assisted creativity, large language models, metacognition, cognitive offloading, creativity suppression, human-AI collaboration, cognitive load theory, expertise development
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-11.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-11.
Cited sources are checked by gpt-6-sol.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
