Abstract
The integration of generative AI into advertising education raises questions about its impact on creative learning. Guided by Amabile’s componential theory, this mixed-methods study compares human–AI and human–human collaboration in ideation tasks completed by 64 undergraduates. AI collaboration increased ideational efficiency but was associated with lower reported cognitive engagement and creative ownership; students viewed AI-generated ideas as more original yet less relevant. Preference for AI collaboration was predicted by perceived creativity during the task rather than baseline creative confidence. These findings suggest that structured sequencing and framing of AI use are critical for sustaining creative agency in advertising coursework.
Keywords
Introduction
Advertising educators face a new reality: students have constant access to generative AI tools like ChatGPT that can produce creative concepts and persuasive copy instantly. This raises urgent pedagogical questions about how we teach creativity when AI can generate ideas immediately (Ford et al., 2023; Leung, 2025). The challenge extends beyond academic integrity to the core of advertising education: if students can outsource ideation to algorithms, how do we cultivate the strategic thinking, cultural sensitivity, and creative judgment that define effective advertising practice?
Answering these questions requires more than tracking what students produce; it requires examining the psychological conditions under which creativity develops. When AI generates ideas on demand, it may alter not just the content of student work but students’ cognitive engagement with the process, their sense of authorship over the output, and their motivation to persist through difficulty. These are not peripheral concerns; they are the foundations on which advertising expertise is built. To examine how AI collaboration shapes these conditions, this study draws on Amabile’s (1996) componential theory of creativity, which identifies domain-relevant skills, creativity-relevant processes, and intrinsic task motivation as the essential psychological foundations of creative performance.
Our study explores these dynamics through an empirical investigation of human–AI versus human–human student collaboration on problem-solving tasks intentionally structured to mirror early-stage advertising ideation. We offer advertising educators empirically grounded guidance on when and how AI integration can support, rather than undermine, creative development.
Theoretical Foundations
To examine how AI integration affects creative development in advertising education, we draw on Amabile’s (1996) componential theory of creativity as our organizing framework. Amabile identifies three essential components of creative performance: domain-relevant skills, creativity-relevant processes, and intrinsic task motivation. This framework allows us to assess whether AI use supports or disrupts the psychological foundations necessary for creative learning. In this study, we interpret domain-relevant skills through students’ opportunities to rehearse advertising thinking, creativity-relevant processes through their reported elaboration, reflection, and fixation, and intrinsic task motivation through their perceived engagement, ownership, and creative confidence.
Domain-Relevant Skills
Domain-relevant skills refer to the knowledge and strategic expertise specific to advertising, including creative thinking, audience analysis, persuasion principles, cultural awareness, and message development. Creative thinking is not supplementary to advertising expertise; it is central to it, shaping how practitioners generate concepts, frame problems, and develop resonant messages. While generative AI may provide rapid access to structured frameworks and idea variations, reliance on AI-generated suggestions may reduce opportunities for deliberate rehearsal of these skills.
Creativity-Relevant Processes
Creativity-relevant processes include divergent thinking, elaboration, iteration, and problem reframing. AI systems can increase ideational fluency during early brainstorming (Wu et al., 2021), but this output does not necessarily translate into deeper elaboration or sustained engagement. If AI suggestions anchor student thinking or reduce cognitive effort, key creativity-relevant processes may be weakened. In this study, we examine these processes through students’ reports of idea generation, elaboration, and anchoring when collaborating with human versus AI partners.
Intrinsic Task Motivation
Intrinsic task motivation refers to ownership, engagement, and personal investment in creative work. AI tools may reduce initial barriers to ideation for some students, yet they may also diminish ownership when ideas are perceived as originating from the system rather than the individual. In our design, these motivational dynamics are reflected in students’ reported engagement, partner preferences, and qualitative descriptions of feeling more or less creative across collaboration modes.
Taken together, Amabile’s framework provides the lens through which we interpret AI integration in this study: we compare human–human and human–AI collaboration to examine how each mode supports or disrupts domain-relevant skills, creativity-relevant processes, and intrinsic task motivation in classroom creative tasks.
Literature Review
Practical approaches to human–AI collaboration in advertising contexts have evolved significantly in recent years. The Human-in-the-Loop (HITL) approach positions AI as a generator of multiple possibilities, iterations, and refinements, with humans serving as curators and directors through prompt engineering and selective implementation (Candy, 2019). This model aligns closely with traditional advertising creative processes, where teams generate, select, and refine concepts through multiple iterations before final execution.
Empirical studies examining human–AI creative partnerships reveal nuanced dynamics. Students and professionals often perceive AI as valuable for generating ideational volume and diversity, particularly during early brainstorming (Ford et al., 2023). However, concerns persist that AI use may diminish critical thinking, reduce creative ownership, and undermine the development of core advertising skills (Habib et al., 2024; Leung, 2025). Collaborative dynamics between humans and AI have also been characterized as fundamentally asymmetrical: AI functions as a divergent stimulus generating output volume, while humans serve as evaluators and selectors rather than co-constructors (Vinchon et al., 2023). This asymmetry has implications for how students develop creative agency, since the evaluative role requires different cognitive engagement than generative participation. These tensions emphasize the importance of understanding how to integrate AI in ways that augment, rather than replace, the creative capacities that advertising education aims to develop.
The most effective implementations of AI in the advertising process position it as a brainstorming partner or ideation aid, rather than a fully autonomous creator (Tang et al., 2025). Student-centered research supports this view. From a student perspective, AI tools are perceived as expanding creative possibilities while simultaneously reducing felt creativity and personal investment in the work (Creely & Blannin, 2025; Marrone et al., 2022). These findings suggest that the augmentation-automation distinction is experienced psychologically, not just operationally. This approach aligns with the broader trend toward augmentation rather than automation, where technology enhances human capabilities without eroding the distinctly human elements of creativity. As these collaborative approaches evolve, advertising practitioners face both opportunities and challenges.
The potential benefits include expanded conceptual exploration, accelerated ideation cycles, and enhanced productivity in content generation (Crimaldi & Leonelli, 2023). However, significant concerns include diminished creative ownership, potential skill erosion, and questions about evaluating the quality and originality of collaboratively produced content. These tensions emphasize the need for intentional approaches to AI integration that leverage technological capabilities while preserving the human judgment, cultural sensitivity, and strategic alignment essential to effective advertising.
Taken together, this body of work points to a consistent tension: AI expands ideational volume while simultaneously narrowing cognitive engagement, ownership, and strategic refinement. These dynamics map directly onto Amabile’s (1996) three components. Concerns about skill erosion and passive consumption speak to domain-relevant skills; the contrast between AI-driven fluency and human-driven elaboration speaks to creativity-relevant processes; and questions of ownership and authorship speak to intrinsic task motivation. Understanding how AI collaboration affects each of these components—and under what conditions it augments rather than displaces them—is the central problem this study addresses.
Research Questions and Objectives
Despite growing recognition of AI’s impact on creativity, empirical research examining human–AI collaboration in advertising education remains limited. While theoretical frameworks have been proposed (Beghetto, 2023; Glăveanu, 2023), pedagogical validation of these frameworks in advertising-specific contexts has been insufficient. Further, while industry reports document increasing AI adoption in advertising practice (Ford et al., 2023; Hartmann et al., 2025), less is known about how these technological integrations influence practitioners’ creative processes, self-perceptions, and attitudes toward collaborative work. Guided by Amabile’s componential theory, this study examines three questions. Amabile’s framework treats intrinsic motivation and process engagement as conditions that must be earned through instructional design, not assumed. Research Question 3 follows from that premise.
Methodology
We employed a within-subjects design comparing human–human and human–AI collaboration. There were 64 undergraduate students in a Social Media Planning course at a U.S. R1 university. Each completed two scenario-based creative tasks: one with a human partner, one with ChatGPT-4o. Students were randomly assigned to begin with either condition to mitigate order effects. Following both tasks, students completed surveys assessing perceived creativity, partner preferences, and creative confidence.
Scenario-Based Problem-Solving Tasks
Each task involved 10 min of brainstorming followed by a Qualtrics survey assessing perceived challenges and strengths. The prompts were designed to be functionally equivalent in their creative and cognitive demands. They differed in surface context (“redesign classrooms” vs. “redesign public spaces”) to reduce carryover effects in the within-subjects design. Although the prompts referenced physical learning and public environments, they were designed to be structurally parallel to early-stage advertising ideation, requiring students to define user needs, generate conceptual “platforms,” and evaluate originality and relevance under time constraints. Both tasks were open-ended and future-oriented, emphasizing ideation and conceptual framing rather than domain-specific expertise.
This design choice was deliberate. A within-subjects study in which both tasks draw on advertising-specific knowledge risks knowledge transfer: ideas, framings, and strategies generated in the first condition carry over and contaminate ideation in the second, making it hard to isolate the effect of the type of collaboration. Functionally equivalent but surface-distinct prompts prevent this without sacrificing comparability. Importantly, the constructs this study measures, ownership, elaboration, cognitive engagement, and anchoring, are process-level and motivational, not domain-content outcomes. General ideation tasks are appropriate for isolating these dynamics because they activate the same creativity-relevant processes and intrinsic motivational conditions that Amabile’s framework predicts will be affected by AI collaboration, regardless of topic.
The pedagogical implications drawn from these findings apply to advertising education precisely because the psychological mechanisms at work are the same ones students bring to campaign development, copywriting, and strategic communication. Students worked in randomly assigned pairs and completed surveys individually. Order of conditions was counterbalanced, and AI familiarity was included as a covariate in regression analyses predicting partner preference.
Human–Human Collaboration
Prompt: “How can we redesign classrooms to enhance creativity?”
Students worked in pairs discussing solutions, then individually completed surveys.
Human–AI Collaboration
Prompt: “How can we redesign public spaces to enhance creativity?”
After a brief ChatGPT tutorial, students brainstormed with AI, then completed surveys.
Students then responded to six open-ended questions probing how human versus AI partners influenced their ideation process, what they learned about AI as a creative partner, and their evolving conceptions of creativity. This classroom-based research design prioritized ecological validity, examining AI collaboration dynamics as they naturally occur in advertising coursework.
Measures
AI Tool Experience
Students reported familiarity with ChatGPT on a 5-point Likert-type scale. Familiarity was high (M = 4.16, SD = 0.80), with 56.3% agreeing and 32.8% strongly agreeing they were familiar with AI tools. Although this item served as a proxy for prior AI exposure, it does not capture more nuanced attitudes or biases toward AI (e.g., enthusiasm, skepticism, perceived threat).
Creative Self-Beliefs
Students completed the Short Scale of Creative Self (SSCS; Karwowski et al., 2018), a psychometrically validated instrument measuring two facets of creative self-perception: creative self-efficacy (CSE) and creative personal identity (CPI). The SSCS comprises 11 items rated on a 5-point Likert-type scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). Six items assessed students’ confidence in their creative problem-solving abilities (CSE), including “I know I can efficiently solve even complicated problems” and “I trust my creative abilities.” Five additional items measured the centrality of creativity to students’ identity (CPI), such as “Creativity is an important part of myself.” The scale demonstrated high internal consistency in this sample (α = .90, M = 4.18, SD = 0.60).
Perceived Creativity and Partner Evaluation
Following each session, students rated agreement with “I felt more creative while working with this partner compared to when I work alone” (1 = strongly disagree to 5 = strongly agree). Human partner ratings (M = 3.94, SD = 1.18) exceeded AI ratings (M = 3.54, SD = 1.19). Students then selected their preferred partner and explained their choice via an open-ended response. Preferences were nearly evenly split (53.1% AI, 46.9% human).
Together, these measures correspond to two of Amabile’s three components. Creative self-efficacy and creative personal identity, as captured by the SSCS, reflect the motivational preconditions for intrinsic task motivation, specifically students’ baseline sense of creative capability and identification with creativity prior to collaboration. Perceived creativity during each session and partner preference address the motivational and process dimensions by indicating how collaboration type shaped felt engagement and ownership in the moment. Domain-relevant skill rehearsal was not directly operationalized through a quantitative measure; it is assessed through the qualitative themes and acknowledged in the Limitations as a priority for future measurement.
Qualitative Analysis
We conducted reflexive thematic analysis (Braun & Clarke, 2019) of open-ended responses (n = 64). Coding was initially inductive, allowing patterns to emerge from student reflections. Once themes were identified, we interpreted them through Amabile’s (1996) componential theory, examining how collaboration types influenced domain-relevant skills, creativity-relevant processes, and intrinsic task motivation. Through iterative coding and refinement, we identified five central themes: (a) Speed and Efficiency, (b) Idea Generation & Creativity, (c) Critical Thinking & Learning Process, (d) AI as Tool vs. Replacement, and (e) Self-perception of Creativity. Theme frequencies were compiled into Table 1.
Summary Table of Emergent Themes.
Note. N = 64 students. Some responses contained multiple themes, so the total count exceeds the number of respondents.
Mixed-Methods Integration
This convergent mixed-methods design (Creswell & Plano Clark, 2023) integrated qualitative and quantitative data during interpretation. Thematic analysis identified experiential patterns, while quantitative analyses tested whether perceived differences were statistically reliable.
Results
The following section presents findings from qualitative and quantitative analyses examining how human–human and human–AI collaborations shaped creative behaviors, perceptions of idea quality, and self-assessed creative confidence among strategic communication students. Drawing on thematic patterns and statistical comparisons, we address each research question in turn to illuminate the distinct affordances and frictions of each collaboration mode, particularly as they relate to creative development and campaign ideation in an AI-enhanced advertising environment. All quantitative analyses report test statistics, effect sizes, and confidence intervals where appropriate to facilitate assessment of robustness and practical significance.
The presentation of results also reflects this framework. Themes emerging from participant reflections were interpreted in relation to the three components of Amabile’s model, allowing us to compare how human–human and human–AI collaborations differentially influenced creative skills, processes, and motivational engagement.
RQ1: What Collaborative Behaviors Emerge When Students Work With Human Partners Versus AI Tools in Advertising-Oriented Creative Tasks?
Qualitative findings revealed distinct behavioral patterns between human–human and human–AI collaborations. Human partnerships were consistently described as reciprocal and co-constructive, with 26 students emphasizing collaborative idea-building: “We bounced ideas off each other and built on each other’s thoughts.” This mode of collaboration appeared to foster active participation, deeper engagement, and mutual refinement of ideas. Several students also highlighted the motivational benefits of working with another person, describing a sense of social accountability: “I wanted to make sure I was contributing to the discussion in a meaningful way.”
In contrast, collaborations with AI were more efficient but unidirectional. The most prevalent theme in participant reflections was the perception that AI tools provided a broader or more novel idea set (n = 34), often described as offering “really good solutions” and “ideas I would not have thought of alone.” However, this perceived benefit came with trade-offs. Twenty-two students said that using AI reduced their need for active thinking or critical engagement. As one explained, “It makes me not work as hard on something and really use my critical thinking skills.” Another reflected, “I just copied and pasted without thinking.”
Some students who began their process with AI reported that those early ideas carried over and constrained later ideation with human partners: “After using AI first, I was thinking more about what it had suggested than coming up with my own ideas.” This mirrors findings from recent studies (e.g., Vinchon et al., 2023) that characterize human–AI collaboration as asymmetrical, with AI functioning more as a divergent stimulus than as a responsive creative partner.
Quantitative results supported these themes, demonstrating that the sense of deeper engagement and cognitive ownership students described in human collaboration corresponded with measurably higher perceived creativity. Students who began with a human partner reported significantly greater feelings of creativity when collaborating (M = 4.40, SD = 0.76) than those who began with AI (M = 3.63, SD = 1.30), t(60.36) = 2.95, p = .005. This difference represented a medium-to-large effect size (Hedges’ g = 0.68, 95% CI [0.16, 1.19]), indicating a meaningful boost in perceived creativity when human collaboration came first. By contrast, students’ perceived creativity when working with AI did not differ significantly by condition order (human-first: M = 3.36, SD = 1.29; AI-first: M = 3.67, SD = 1.13; t(46.47) = –0.97, p = .335).
A paired-samples t-test showed a similar pattern, though the difference was not statistically significant. Students reported feeling slightly more creative when working with a human partner (M = 3.94, SD = 1.18) than with AI (M = 3.54, SD = 1.19), t(62) = 1.89, p = .064, which aligns with qualitative accounts suggesting that human–human collaboration fosters more active engagement and creative agency than AI-supported ideation.
In sum, RQ1 findings reveal that while AI collaboration streamlines ideation and introduces novel concepts, human collaboration fosters deeper engagement, co-construction, and a stronger sense of cognitive ownership, qualities central to creativity-relevant processes as conceptualized by Amabile (1996).
RQ2: How Does Collaboration Type (Human–Human vs. Human–AI) Influence Students’ Perceptions of Creative Output Quality, Including Originality, Relevance, and Motivational Engagement?
Students’ assessments of output quality revealed meaningful trade-offs across collaboration types. AI-generated contributions were widely praised for their originality and ideational breadth. Thirty-four students characterized AI ideas as “unique,” “outside-the-box,” or “different from what I would have thought of.” However, this novelty was often tempered by concerns about practicality, contextual fit, or ownership. Eighteen students reported feeling detached from the output, with one noting, “I wasn’t able to build on any of them and felt as if I was just copying down answers.”
By contrast, ideas generated through human collaboration were more frequently described as relevant, grounded, and strategically usable. Nineteen students said their team-developed concepts were “more realistic,” “more effective,” or “more practical.” One reflected, “The AI was faster, but working with humans made the ideas better and more usable.” Students also articulated a nuanced view of AI’s role in professional creative work. Students expressed cautious enthusiasm, generally viewing AI as a starting point rather than a replacement for human creativity.
Quantitative findings supported this pattern of cautious enthusiasm. A 2 (Modality: Human vs. AI; within-subjects) × 2 (Group: Human-First vs. AI-First; between-subjects) mixed ANOVA revealed no significant main effect of modality, F(1, 47) = 1.67, p = .203, partial η² = .03, and no significant main effect of group, F(1, 47) = 2.07, p = .157, partial η² = .04. The Modality × Group interaction was not significant, F(1, 47) = 3.30, p = .076, partial η² = .07, though the pattern suggested a possible order effect that may warrant further investigation in larger samples. Because SPSS uses listwise deletion for repeated-measures GLM, this analysis included n = 49 participants with complete data on all three variables. In addition, perceived creativity during AI collaboration was moderately correlated with partner preference, r = .35, p = .004, indicating that novelty likely influenced students’ satisfaction with AI, even when the ideas themselves felt less actionable. Familiarity with AI tools was included as a covariate in all regression analyses predicting partner preference to account for potential novelty effects associated with generative AI use. While AI familiarity did not predict perceived creativity outcomes, it emerged as a significant predictor of preference for AI collaboration.
A logistic regression predicting partner preference from creative self-confidence, AI familiarity, perceived creativity with a human partner, perceived creativity with AI, and group assignment was significant overall, χ²(5) = 17.87, p = .003, Nagelkerke R² = .33, correctly classifying 73.0% of cases (N = 63). Feeling more creative during AI collaboration predicted preference for AI (OR = 2.08, 95% CI [1.22, 3.55], p = .007), as did familiarity with ChatGPT (OR = 2.44, 95% CI [1.01, 5.94], p = .049). Creative self-confidence did not predict preference (OR = 1.02, 95% CI [0.36, 2.89], p = .974), nor did perceived creativity with a human partner (OR = 0.71, 95% CI [0.39, 1.31], p = .277) or group assignment (OR = 2.46, 95% CI [0.72, 8.43], p = .152).
A linear regression treating AI preference as a continuous rating was also significant, F(3, 60) = 2.85, p = .045, R² = .13, adjusted R² = .08. AI familiarity emerged as a significant predictor (B = 0.17, SE = 0.08, β = 0.28, p = .030), group order was not statistically significant (B = 0.24, SE = 0.12, β = 0.24, p = .057), and creative self-confidence did not predict preference (B = –0.03, SE = 0.10, β = –0.03, p = .803). These findings reinforce the idea that experiential novelty, rather than stable traits, most influences perceptions of AI’s collaborative value.
Although preference for collaboration was nearly evenly split (53.1% favored AI, 46.9% favored human), those who began in the AI condition reported slightly stronger AI preference ratings (M = 4.62, SD = 0.49) than those who started with a human partner (M = 4.40, SD = 0.50). This difference did not reach statistical significance, t(50.77) = −1.69, p = .097, and no reliable conclusions about the effect of condition order on preference strength can be drawn from this comparison.
RQ2 findings suggest that although AI-generated ideas were seen as more original, human collaborations produced outputs perceived as more usable, relevant, and personally meaningful, highlighting a potential tension between divergent ideation and strategic refinement within the creativity-relevant process component.
RQ3: Do Collaborations With AI Require Distinct Pedagogical Strategies to Preserve Students’ Cognitive Engagement and Creative Agency Compared to Human–Human Creative Partnerships?
Where RQ2 examined how the type of collaboration shaped perceptions of output quality, RQ3 addresses the process and motivational experiences of collaboration itself. Specifically, whether AI-assisted ideation sustains the cognitive engagement and creative agency central to Amabile’s framework.
Participant responses surfaced clear tensions in how AI is currently perceived within creative workflows. While many students acknowledged the efficiency and expansive idea generation offered by tools like ChatGPT, there was widespread consensus that AI should serve as a generative aid rather than a replacement for human creativity. A strong majority (n = 31) believed AI should augment rather than replace human creativity. Many (n = 27) framed AI tools as useful for jumpstarting idea generation but not suitable as stand-alone creative engines: “It can help get the ball rolling initially, but it really isn’t that creative on its own.” This perspective reflects a growing recognition that while AI tools can increase output volume and accelerate timelines, they do not yet replicate the iterative, interpretive, or emotionally resonant dimensions of human creative labor.
Concerns about overreliance on AI tools also emerged, pointing to anxieties about the erosion of creative instinct and strategic thinking over time. Fifteen students expressed fears of skill atrophy in relation to repeated AI use, with one noting, “If you keep using it, your own resources will diminish.” Fourteen others indicated that using AI diminished their felt sense of creativity, with comments such as, “I lost all of my touch . . . even when it came to a simple question, I had nothing in my brain.” These are student-reported perceptions of AI’s effect on felt creativity and creative instinct; whether analogous patterns occur among professional advertising practitioners remains an open question for future research.
A related insight involved how students experienced the interpersonal dynamics of collaboration. When working with AI, students overwhelmingly described the process as unidirectional and isolating. One respondent stated, “I just copied and pasted without thinking,” contrasting this experience with human collaboration, which they described as more engaging and mutually responsive. Seventeen students specifically emphasized the importance of interpersonal connection, conversation, and real-time feedback in creative ideation. One participant remarked, “I preferred working with a partner because we actually discussed our thoughts . . . with AI, it felt one-sided.” These findings emphasize the affective and dialogic dimensions of creative development, factors that remain difficult for AI to simulate meaningfully.
Moving from subjective experiences to measured self-beliefs, students’ creative self-confidence was high overall (M = 4.18, SD = 0.60, α = .90). Scores did not significantly differ based on collaboration order, t(62) = –0.29, p = .77. Creative confidence was not significantly associated with perceived creativity during either human collaboration, r = .20, p = .110, or AI collaboration, r = –.08, p = .518. Consistent with the regression findings, this suggests that the stable trait of creative self-confidence does not translate directly into in-the-moment creativity during collaboration; rather, felt creativity is shaped by the interaction itself.
Notably, although familiarity with AI predicted preference in regression analyses, its effect was smaller than that of in-the-moment perceived creativity during AI collaboration, suggesting that enthusiasm for AI collaboration is shaped both by prior exposure and, more substantially, by immediate experience. Together, these findings suggest that while AI tools may offer useful ideational support, they may not foster the same sense of confidence, engagement, or ownership that emerges through human collaboration. These patterns suggest that the motivational and relational conditions Amabile identifies as central to creative development are not reliably activated by AI collaboration, which raises direct implications for how AI use is structured in creative coursework.
Discussion
This study explored how human collaboration with AI affects creative ideation processes, perceptions of creative agency, and self-reported creative confidence among advertising students. The psychological patterns documented here (reduced cognitive engagement, ownership erosion, and anchoring on AI-generated suggestions) are the same processes at work when advertising students develop campaign concepts, write copy, or conduct audience analysis with AI assistance. The advertising education context is not incidental; it is the frame through which these mechanisms carry practical significance.
Anchored in Amabile’s (1996) componential theory of creativity, the findings illuminate how AI reshapes domain-relevant skills, creativity-relevant processes, and intrinsic task motivation, three pillars essential to persuasive message development. Rather than merely serving as tools, AI systems emerged as active agents that altered the content of creative output and students’ cognitive and affective orientation toward idea generation. To synthesize these findings, we interpret the observed effects through Amabile’s (1996) three components: domain-relevant skills, creativity-relevant processes, and intrinsic task motivation, emphasizing how AI alternately augments and constrains each dimension of creative performance.
Domain-Relevant Skills
Findings suggest that AI tools may expand access to idea fluency and conceptual variation, but may also reduce opportunities for deliberate rehearsal of advertising expertise when students rely on AI-generated frameworks. This framing of AI as a “creative partner” aligns with growing concerns about the intentionality users assign to non-human agents (Huh et al., 2023).
For advertising educators, these findings show that AI’s influence extends beyond the content students produce to how they perceive their own creative agency. When students attribute idea generation primarily to AI, they experience reduced ownership, even when they have made strategic decisions about which AI suggestions to use. This suggests that assignments must explicitly position students as creative directors who evaluate and refine AI outputs rather than as consumers selecting from algorithmic options. Allowing the use of AI without structured, critical engagement risks students developing passive relationships with these tools, undermining the creative confidence and strategic judgment central to advertising expertise.
Creativity-Relevant Processes
Although AI collaboration broadened perceived originality, it simultaneously undermined creativity-relevant processes such as elaboration, reflection, and strategic refinement. Students described AI-facilitated brainstorming as cognitively passive, “easy” but “mind-numbing,” reinforcing concerns that uncritical reliance on AI may reduce cognitive effort and inhibit the kind of conceptual rigor central to persuasive communication (Habib et al., 2025; Vinchon et al., 2023). Fixation effects further complicated this dynamic: several students reported anchoring their human-led ideation on prior AI-generated suggestions, a pattern that echoes algorithmic anchoring documented in other domains (Ford et al., 2023; Habib et al., 2024). Such effects could compromise creative flexibility and responsiveness, which are critical for tailoring persuasive messages to dynamic contexts.
Motivational responses were similarly complex. While some students described AI as boosting creative confidence by breaking through initial idea blocks, others reported feeling less capable and less creative after using AI. These ambivalent responses suggest that AI shapes not only the output of creative work but also the internal psychological conditions under which creative persuasion occurs (Creely & Blannin, 2025; Marrone et al., 2022).
The results suggest that creative self-efficacy may not only represent an outcome of collaboration but could also influence how individuals approach AI interactions. Students with higher baseline creative confidence may be more likely to use AI critically and selectively, whereas those with lower confidence may defer to algorithmic suggestions more readily. This interpretation aligns with social-cognitive models proposing that efficacy beliefs guide self-regulation and engagement (Bandura, 2001; Tierney & Farmer, 2011). Although moderation was not directly tested in this study, acknowledging this mechanism strengthens the theoretical coherence between self-beliefs and patterns of human–AI collaboration.
Understanding Augmentation Versus Automation in Creative Learning
Interpreted through Amabile’s (1996) framework, findings suggest a distinction between augmentation and automation in how AI affects the three components of creativity. This augmentation-automation distinction, interpreted through Amabile’s (1996) framework, explains when AI supports versus undermines creative development.
Augmentation occurs when AI extends student capability while preserving creative ownership. Students experienced augmentation when they:
Retained interpretive control over which AI suggestions to adopt
Used AI outputs as stimuli for further human elaboration
Applied domain knowledge to evaluate strategic appropriateness
The quantitative results reinforce this interpretive ownership mechanism. Preference for AI collaboration was predicted by feeling more creative during AI use (OR = 2.08, p = .007) rather than by baseline creative self-beliefs, and familiarity with ChatGPT predicted preference without affecting perceived creativity outcomes. Students who experienced AI as augmenting their work tended to report higher felt creativity during collaboration, whereas those who described copying and pasting without thinking also reported lower engagement and reduced ownership. Together, these patterns suggest that it is the subjective experience of agency and engagement during collaboration, rather than stable traits, that most strongly shapes whether AI functions as augmentation or as automation in this context.
In these instances, Amabile’s three components remained intact. Domain-relevant skills were exercised through critical evaluation. Creativity-relevant processes continued through elaboration and refinement. Intrinsic task motivation persisted because students attributed success to their strategic judgment despite AI assistance.
Automation occurs when AI supplants human creative effort. Students experienced automation when they:
Copied AI outputs without reflection
Deferred strategic judgment to algorithmic suggestions
Allowed AI-generated ideas to anchor subsequent thinking
In these instances, students described patterns consistent with reduced engagement in all three components. Opportunities for independent rehearsal of domain-relevant skills appeared reduced. Creativity-relevant processes (elaboration, iteration) disappeared. Students reported diminished ownership, suggesting possible implications for intrinsic task motivation.
The mechanism distinguishing these modes is preservation of interpretive ownership. When students maintain active engagement by questioning AI outputs, applying advertising knowledge to assess strategic fit, and refining suggestions, AI appears to augment their creative capacity. When ownership erodes through uncritical acceptance, the same technology automates away the cognitive work essential to creative development.
For advertising educators, this distinction provides a lens for designing assignments. When students function as creative directors evaluating AI’s work, learning occurs. When students function as consumers selecting from algorithmic menus, creative development suffers. The difference lies not in the AI tool itself but in how we structure students’ engagement with it.
Implications for Advertising Education
The augmentation-automation distinction offers practical guidance for integrating AI into advertising curricula. Effective integration requires pedagogical strategies that preserve students’ interpretive control and creative ownership.
Require Human-First Ideation
Students should develop initial campaign concepts, strategic frameworks, or brand messaging independently before consulting AI tools. This sequencing ensures that AI refines rather than generates strategic thinking. Our data showed that students who began with AI experienced diminished creative confidence and relied on algorithmic suggestions. Requiring human-first work protects the strategic foundation that AI cannot replicate.
Build Domain Knowledge for Critical Evaluation
Although our study did not directly measure domain knowledge, prior work and Amabile’s framework suggest that AI’s augmenting potential depends on students possessing expertise to evaluate outputs for strategic fit, cultural appropriateness, and brand alignment.
Embed Structured Reflection
AI collaboration often produces cognitive passivity. To counteract this, require reflection after AI-assisted work: What surprised you about AI’s suggestions? Where did you disagree and why? What strategic insights did you contribute that AI could not? Reflection reinforces ownership and helps students articulate their creative value beyond idea generation. This recommendation responds to students’ descriptions of AI sessions where they “just copied and pasted without thinking,” and their reports that such passive collaboration left them feeling less creative and less connected to the work.
Align AI Use With Creative Process Stages
AI contributes most effectively during early divergent ideation, where rapid exploration expands possibilities. In our sample, many students (n = 27) described AI as useful for “jumpstarting” or “getting the ball rolling” in brainstorming but not as a stand-alone creative engine, which supports positioning AI primarily in early exploration rather than in later evaluative phases. In later convergent phases requiring judgment, contextual alignment, and narrative coherence, AI often produces superficial solutions. Strategic timing—deploying AI for exploration while reserving evaluation and refinement for human expertise—maximizes augmentation while minimizing automation risks.
Scaffold AI Integration Developmentally
Although our study did not manipulate course level or longitudinal exposure, the patterns we observed suggest a possible value in a graduated approach to AI integration. One theoretically grounded sequence would restrict AI in foundational courses to build creative confidence, require AI consultation with mandatory evaluation in intermediate courses, and treat AI as optional in advanced courses, with students justifying their choices.
Assessment Approaches
Traditional output-only rubrics become insufficient when AI can generate polished work. Assessment must evaluate strategic thinking and process alongside products. Effective criteria include: evidence of independent strategic analysis, quality of AI output evaluation, strategic justification for adopting/rejecting suggestions, demonstration of cultural knowledge AI cannot access, and originality of synthesis. Process documentation—annotated work samples, creative journals, and think-aloud protocols—becomes essential for evaluating learning.
Course Policies
Transparent communication about when AI use is encouraged, restricted, or optional prevents confusion. Foundational courses benefit from AI restrictions that foster creative confidence. Intermediate courses can require AI consultation with a mandatory evaluation. Advanced courses treat AI as one optional tool among many, with students justifying their choices. The goal is to develop capabilities that AI cannot replace, not to prevent AI use.
Conclusion
This classroom-based study provides advertising educators with empirical evidence about how AI collaboration affects student creative development. Our findings reveal that AI’s impact on learning depends fundamentally on how we structure its use. When students maintain critical judgment and strategic control over AI outputs, collaboration can expand creative exploration. When students passively consume algorithmic suggestions, the same technology undermines cognitive engagement, creative ownership, and skill development—the capacities advertising education aims to cultivate.
As AI capabilities advance, advertising programs must integrate these technologies deliberately. Students need preparation to use AI critically—as creative directors evaluating options, not consumers selecting from menus. This requires assignment design positioning students as decision-makers, instruction in evaluating outputs, and assessment rewarding strategic thinking.
Future research should examine how AI-integrated pedagogy affects long-term professional readiness, assessment effectiveness across contexts, and creative self-efficacy development. As advertising educators, we face not a choice between embracing or rejecting AI, but a responsibility to teach students when, how, and why to engage it thoughtfully. This study demonstrates that with intentional pedagogical design, we can prepare students for AI-integrated professional practice while preserving the creative development fundamental to advertising expertise.
Limitations
Several limitations inform the interpretation of these findings. This study examined student experiences in short-term, low-stakes creative tasks embedded in a single course at one institution. While this design prioritized ecological validity, findings should be validated across diverse institutional settings, course levels, and student populations. Advanced students with deeper advertising expertise may engage with AI differently from undergraduates in our study, and dynamics may shift in higher-stakes contexts such as capstone portfolio courses or agency-client projects.
Our within-subjects design comparing human–human and human–AI collaboration captured immediate differences but could not assess long-term impacts on skill development, creative confidence, or professional readiness. Longitudinal research examining how repeated use of AI affects creative capacity over months or years would strengthen understanding of the skill-atrophy concerns that students articulated. Similarly, our scenario-based tasks, while advertising-oriented, cannot fully replicate the complexity of multi-week campaign development, which involves client feedback, team dynamics, and iterative refinement.
In addition, the study measured AI familiarity as a covariate but did not assess participants’ pre-existing attitudes toward AI. Dispositional orientations toward AI may shape engagement behavior and self-reported outcomes independently of prior experience, and their absence as a measured variable is a limitation. Future research incorporating validated AI attitude scales would allow for a more precise account of how beliefs about AI mediate creative collaboration dynamics.
The study used ChatGPT-4o as the AI collaboration tool, reflecting current widespread availability but limiting generalizability to other AI systems with different capabilities, interaction styles, or training data. As AI technologies evolve rapidly, findings may shift with more sophisticated tools. In addition, we did not manipulate or control specific modes of AI engagement (e.g., AI as an idea generator, critic, or refiner), leaving open questions about how different functional roles for AI might produce distinct motivational and creative effects.
The use of general ideation prompts rather than advertising-specific briefs also limits direct generalization to domain-intensive tasks such as brand strategy, creative concepting, or consumer insight development. Whether the motivational and process-level patterns observed here hold when domain knowledge is actively required remains an open question. Future research using advertising-specific scenarios would allow a more direct test of AI’s effects on domain-relevant skill rehearsal as Amabile conceptualizes it.
Finally, this exploratory study focused on students’ experiences and self-perceptions rather than on external evaluations of creative output quality. While student perceptions matter for understanding motivation and engagement, future research incorporating expert assessments of work produced through human–AI collaboration would provide complementary evidence on pedagogical effectiveness. Intrinsic task motivation was inferred from qualitative reflections rather than directly measured using a validated motivation scale. Similarly, domain-relevant skill acquisition was not assessed longitudinally. Future research should incorporate direct measures of these components to strengthen causal inference. Despite these limitations, this study provides empirically grounded insights into how AI collaboration affects the creative development that advertising educators aim to cultivate.
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.