Reclaiming Creativity in the Age of AI Learning

Author: Denis Avetisyan

As artificial intelligence reshapes education, understanding how students maintain ownership of their creative process is more crucial than ever.

This review explores the relationship between student agency, agentic engagement, and the development of ‘mini-c’ creativity within AI-assisted learning environments.

While traditionally framed as outputs of individual cognition, creativity is increasingly shaped by interactions within complex learning systems. This paper, ‘Human Agency and Creativity in AI-Assisted Learning Environments’, examines how student agency-specifically through a proposed framework of ‘agentic engagement’-mediates creative processes when generative AI tools are integrated into learning. We argue that fostering this agentic engagement shifts the focus from creative products to a process of personally relevant meaning-making, linked to the concept of ‘mini-c creativity’. How can we best design AI-assisted learning environments to cultivate, rather than constrain, students’ agentive participation and unlock their creative potential?

The Shifting Sands of Agency

Conventional understandings of agency, particularly those exemplified by Instrumental Agency, frequently prioritize the successful attainment of predetermined goals within a learning context. However, this narrow focus can inadvertently obscure the intricate and often messy processes of learning itself. A student’s journey isn’t solely defined by reaching an end point; it’s characterized by exploration, adaptation, and the negotiation of challenges. The emphasis on outcomes, while important, risks overlooking the cognitive and emotional work involved in sense-making, the development of understanding, and the construction of knowledge-essential elements that contribute to a deeper, more meaningful learning experience. Consequently, a purely goal-oriented perspective may fail to capture the full scope of a student’s agency, neglecting the vital role of intrinsic motivation and the internal shifts that signify genuine learning.

The evolving landscape of education, increasingly interwoven with artificial intelligence, demands a re-evaluation of how student agency is understood. Traditional models often frame agency as the ability to initiate action and achieve predefined goals, but this proves insufficient when learners interact with adaptive systems that dynamically reshape the learning path. A holistic perspective recognizes that agency isn’t simply about doing but about navigating complexity, interpreting feedback from AI tutors, and making choices aligned with personal values and evolving understandings. This nuanced view acknowledges that agency is situated – deeply influenced by the specific learning context, the AI’s affordances, and the student’s intrinsic motivation – and requires assessing not just whether a student acts, but how and why, particularly when faced with choices presented by intelligent learning technologies.

Effective learning isn’t solely about reaching a predetermined outcome; instead, it’s deeply interwoven with the individual’s internal drive and the surrounding environment. Current research highlights that motivation-the intrinsic desire to engage with material-plays a crucial role in how knowledge is constructed and retained. Furthermore, the context in which learning occurs – encompassing social interactions, available resources, and even emotional state – significantly shapes the process. Importantly, learners actively synthesize new information with existing beliefs and experiences, forging personal meaning that solidifies understanding and promotes long-term retention. This shift in perspective moves beyond simply measuring achievement toward appreciating the complex interplay between a student’s internal state, external influences, and the construction of knowledge as a uniquely individual endeavor.

The Fuel of Agency: Self-Determination and the Thirst for Mastery

Effortful Agency, as described by Self-Determination Theory (SDT), posits that intrinsic motivation – engagement in an activity for its inherent satisfaction – is fundamentally driven by the fulfillment of three basic psychological needs. These needs are autonomy, the feeling of volition and choice in one’s actions; competence, the perception of mastery and effectiveness in performing tasks; and relatedness, the sense of connection and belonging with others. SDT research demonstrates that environments supporting these needs foster greater intrinsic motivation, leading to increased effort, persistence, and improved performance, while thwarting these needs diminishes motivation and can lead to disengagement.

The fulfillment of psychological needs for autonomy, competence, and relatedness is directly correlated with sustained engagement in learning activities. Research indicates that when individuals perceive control over their learning path – experiencing autonomy – and believe they possess the skills to succeed – demonstrating competence – their intrinsic motivation increases. Furthermore, feeling connected to others and experiencing a sense of belonging – achieving relatedness – reinforces this motivation and fosters a stronger sense of ownership over the learning process. Conversely, the frustration of these needs is associated with decreased motivation, disengagement, and a diminished perception of agency.

Learner investment in education is demonstrably increased when individuals perceive agency through choice, skill development, and social connection. Research indicates that providing options in learning activities supports autonomy, leading to greater intrinsic motivation and sustained engagement. Similarly, opportunities to build competence through mastery-oriented tasks foster a sense of capability and self-efficacy. Finally, establishing supportive relationships and a sense of belonging addresses the need for relatedness, further solidifying learner commitment and promoting active participation in the educational process. These factors collectively contribute to a more profound and enduring investment in learning outcomes.

The Distributed Mind: Agency as an Emergent Property of Interaction

Dynamically Emergent Agency, rooted in Sociocultural Theory, posits that an individual’s capacity to act – their agency – is not an inherent trait but rather a product of ongoing interactions. This framework understands agency as emerging from the reciprocal relationships between the individual, the physical and social environment, and the tools or materials present. Specifically, agency is not solely determined by internal cognitive processes; instead, it’s distributed across these interacting elements. The environment provides affordances – opportunities for action – which are then interpreted and utilized by the individual in conjunction with available materials. This process is iterative and contextual, meaning agency is continuously negotiated and redefined based on the specific situation and the evolving interplay of these components. Consequently, understanding agency requires analyzing the dynamic configuration of these interactions, rather than focusing solely on internal states or individual characteristics.

Learning is fundamentally a social process, meaning knowledge construction is deeply embedded within and influenced by the cultural and collaborative contexts in which it occurs. Individuals do not learn in isolation; rather, they acquire understanding through interactions with others and engagement with culturally-defined tools and practices. These interactions shape not only what is learned, but also how learning takes place, with cultural norms dictating acceptable modes of participation and knowledge sharing. Collaborative experiences, such as peer instruction and group problem-solving, further contribute to this socially situated learning by providing opportunities for individuals to articulate their understanding, receive feedback, and co-construct knowledge with others. The specific cultural values and collaborative structures present within a learning environment, therefore, significantly impact the learning outcomes achieved.

AI-assisted learning environments should prioritize features enabling collaborative knowledge building and responsive adaptation to user interactions. This requires systems that move beyond delivering static content and instead facilitate dialogue, peer-to-peer learning, and the negotiation of meaning. Effective design incorporates mechanisms for shared artifact creation, such as collaborative documents or simulations, and tools for providing and receiving feedback. Furthermore, these environments must be capable of analyzing interaction data to identify patterns in learner behavior and dynamically adjust the learning path or offer personalized support, thereby fostering a continuous cycle of co-construction and refinement of understanding.

The Alchemy of Co-Creation: Amplifying Agency Through Generative AI

Co-creation, facilitated by advances in Generative AI, signifies a fundamental shift in how creativity and agency are cultivated. This approach moves beyond simply utilizing AI as a tool for task completion, instead establishing a collaborative dynamic where humans and artificial intelligence jointly shape outcomes. Effective co-creation hinges on skillful prompt engineering – the art of crafting precise instructions that guide the AI’s generative process – allowing learners to articulate their visions and receive responses that spark further exploration. The result isn’t merely an efficient workflow, but a powerful engine for ideation, where the learner maintains authorial control while leveraging the AI’s capacity for pattern recognition and novel combination. This synergistic relationship fosters a sense of ownership and empowers individuals to actively participate in the creative process, ultimately amplifying their capacity for innovation and self-expression.

The emerging dynamic between humans and artificial intelligence in learning environments centers not on automation, but on amplification. Current research demonstrates that GenAI tools are most effective when positioned as collaborators, offering learners opportunities to expand upon initial concepts, receive iterative feedback on their work, and meticulously refine their outputs. This approach fosters a cycle of exploration and improvement, where AI serves as a responsive partner, providing suggestions and insights without dictating the creative process. The result is a learning experience characterized by increased agency, allowing individuals to maintain ownership of their ideas while benefiting from the computational power and vast knowledge base of AI systems.

This research introduces a theoretical framework demonstrating a strong connection between a learner’s agentic engagement – their proactive and intentional interaction – and the development of mini-c creativity within AI-assisted learning environments. The framework posits that simply providing AI tools isn’t enough; genuine creative growth occurs when students take ownership of their learning process, actively experimenting with prompts, critically evaluating AI-generated outputs, and purposefully refining their ideas. This purposeful interaction, fueled by student initiative, fosters a continuous cycle of exploration and refinement, leading to uniquely personal and novel expressions of understanding – the essence of mini-c creativity. The study emphasizes that the power of generative AI lies not in automating creativity, but in amplifying a student’s existing capacity for it, contingent upon their active and deliberate engagement with the technology.

Toward Authorial Agency: Reclaiming the Narrative of Learning

The concept of authorial agency reframes the learner not as a passive recipient of information, but as an active creator of understanding. This perspective suggests that genuine learning occurs when individuals move beyond simply absorbing facts and instead transform external data into personally meaningful knowledge. By actively shaping their learning path-choosing resources, defining questions, and constructing interpretations-students author their own educational narratives. This process isn’t merely about recall; it’s about building a unique, internal framework where new information connects with existing beliefs and experiences, fostering deeper comprehension and lasting retention. The learner, therefore, becomes the primary architect of their cognitive landscape, building a personalized body of knowledge that extends far beyond the confines of a textbook or classroom.

Agentic engagement represents a fundamental shift in pedagogical approach, positioning students not as passive recipients of information, but as active directors of their own educational experience. This involves cultivating a classroom environment where learners proactively select learning goals, monitor their progress, and adjust strategies based on self-assessment. Such intentionality goes beyond simply completing assigned tasks; it demands that students articulate why they are learning something, how it connects to their existing knowledge, and what steps they will take to deepen their understanding. By fostering this sense of ownership, educators empower students to move beyond rote memorization and cultivate a genuine desire for lifelong learning, driven by personal curiosity and a commitment to self-improvement.

The pursuit of education increasingly prioritizes the development of Mini-c Creativity, a concept denoting the everyday, personally meaningful insights that signal genuine comprehension. This isn’t about fostering prolific artists or groundbreaking innovators, but rather nurturing the capacity for individuals to form unique connections with information, thereby transforming it into personalized knowledge. These ‘small c’ creative acts – a novel application of a concept, a fresh perspective on a familiar problem, or a deeply felt resonance with a new idea – are indicators of true understanding and represent a critical step beyond rote memorization. By emphasizing these personal breakthroughs, education can equip individuals not just with a body of facts, but with the ability to learn, adapt, and innovate throughout their lives, fostering a sense of agency and empowering lifelong intellectual curiosity.

The pursuit of genuinely creative learning environments, as detailed in this exploration of AI-assisted education, reveals a landscape less about control and more about fostering emergent properties. The article’s focus on ‘agentic engagement’ – the capacity of students to shape their learning – echoes a fundamental truth: systems aren’t built, they grow. Ada Lovelace observed, “The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform.” This sentiment, though directed at early computing, remains remarkably prescient. The AI serves not as a creator, but as an extension of the student’s own capacity for ‘mini-c creativity’ – those personal insights that, when nurtured, blossom into something more substantial. Order, in this context, is merely a temporary reprieve – a cache between inevitable shifts in understanding and innovation.

The Looming Shadows

The emphasis on ‘agentic engagement’ feels less like a solution and more like a meticulously documented postponement. This framework, linking student action to nascent creative sparks – ‘mini-c’ creativity – posits a control that systems rarely permit. Each carefully crafted prompt, each algorithmic nudge towards ‘agency,’ implicitly defines the boundaries of permissible insight. The study correctly identifies the shape of the problem – the negotiation between human intention and automated response – but neglects to fully acknowledge the inevitability of erosion. In three releases, the definition of ‘engagement’ will shift, the parameters of ‘agency’ will constrict, and these carefully measured sparks will flicker differently.

The pursuit of co-agency is, at its core, a desire to externalize cognitive load. A belief that shared intention can somehow circumvent the fundamental asymmetry: the system scales, the learner does not. Future work will inevitably focus on quantifying this imbalance – charting the decay of personalized response in the face of generalized models. The real challenge isn’t fostering creativity within the system, but anticipating the forms of creativity that will emerge around it – the adaptations, the workarounds, the beautiful, messy, unpredictable expressions of human ingenuity born from limitation.

This is not a failure of design; it is a consequence of growth. The ecosystem will not be built; it will evolve, and its most interesting features will not be those that were intended, but those that were unforeseen. The study provides a snapshot of a moment before the inevitable shift – a valuable record, but not a prophecy.

Original article: https://arxiv.org/pdf/2512.07117.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/