By Cristo León, Ph.D.
Last reviewed on August 29, 2026.
Generative artificial intelligence has rapidly entered education through a vocabulary of assistance. AI assists students with writing, researchers with literature searches, instructors with instructional materials, and professionals with data analysis. This language remains useful, but it is increasingly insufficient to describe what occurs during sustained interaction with generative AI.
When a person proposes an idea, receives an AI-generated response, evaluates it, rejects alternatives, introduces new constraints, modifies the direction of inquiry, and incorporates unexpected outputs into an emerging intellectual product, the process extends beyond simple assistance.
It begins to resemble co-creation.
AI-aware co-creation is an asymmetrical human–AI process in which generative AI contributes possibilities to iterative intellectual production while humans retain epistemic agency, evaluative authority, situated judgment, and accountability for the resulting knowledge.
From AI Assistance to Human–AI Co-Creation
A conventional tool usually has a relatively predictable relationship with human intention. A calculator executes an operation. A word processor records and manipulates text. A database retrieves information according to specified parameters.
Generative AI complicates this relationship because its outputs can influence the direction of the activity itself.
Imagine a researcher beginning with proposition A. The researcher asks an AI system to interrogate that proposition. The system generates alternatives B and C. The researcher rejects B, modifies C, and introduces contextual knowledge that leads to D. The AI then identifies a possible relationship between C and D that the researcher had not previously considered. After evaluating that relationship, the researcher reformulates the original proposition as E.
The resulting intellectual process is no longer adequately represented as:
Human → Tool → Product
Instead, it more closely resembles:
Human ↔ AI → Evaluation ↔ Reformulation → Emergent Product
This does not mean that humans and artificial intelligence possess equivalent agency, intentionality, authorship, or responsibility. They do not.
Rather, it means that the process through which an intellectual object develops can become iterative, relational, and partially emergent. This interactional structure is where the concept of co-creation becomes analytically useful.
What Is Asymmetrical Co-Creation?
Asymmetrical co-creation describes human–AI collaboration in which both human and artificial contributions can influence an emerging intellectual product without assuming equivalence in agency, intentionality, authorship, or responsibility.
This distinction is necessary because describing AI as a co-creator can easily lead to anthropomorphism.
A large language model does not participate in intellectual work with the same lived experience, disciplinary identity, intentions, ethical commitments, or accountability as a human participant. Human–AI co-creation should therefore not be understood as symmetrical collaboration.
The Human Role in Co-Creation
The human participant establishes purposes, formulates questions, introduces contextual and disciplinary knowledge, evaluates relevance, rejects inappropriate outputs, determines when the process should change direction, and remains accountable for the resulting intellectual product.
The AI Role in Co-Creation
The generative AI system contributes variability. It can produce alternative formulations, associations, transformations, simulations, critiques, recombinations, and possible intellectual trajectories.
Co-creation therefore describes the interactional structure of production, not equality between the participants.
Co-Creation Requires Epistemic Agency
The distinction between assistance, delegation, and co-creation becomes particularly important in education.
If a student asks a generative AI system to produce an essay and submits that text without meaningful evaluation or transformation, the student has not engaged in substantial co-creation. The dominant process is delegated production.
Now consider a student who develops an argument, uses AI to challenge it, compares alternative explanations, verifies evidence, rejects unsupported propositions, reformulates the argument, documents AI contributions, and assumes responsibility for the final claim.
This represents a fundamentally different intellectual process.
The critical variable is therefore not simply the amount of AI use. It is the student’s exercise of epistemic agency.
Epistemic agency, in the context of human–AI interaction, is the capacity to determine what intellectual work should be performed personally, collaboratively, or delegated to an artificial system while retaining responsibility for evaluating the resulting knowledge claims.
This produces a useful conceptual sequence:
AI awareness → Epistemic agency → Epistemic delegation → Intellectual effort → Co-creation → Accountability
Co-creation without evaluation can become dependence. Delegation without awareness can become substitution. Generation without accountability can undermine authorship and the credibility of knowledge production.
The educational objective should therefore not be maximum AI use or minimum AI use. It should be the development of learners capable of making intentional decisions about the distribution of intellectual effort.
AI Literacy Is Not Enough
AI literacy is increasingly presented as a necessary educational outcome. However, knowing how to operate an AI system is only one component of responsible human–AI interaction.
A useful distinction can be made among several related concepts:
- AI literacy concerns capability.
- AI awareness concerns context.
- Epistemic agency concerns choice.
- Epistemic delegation concerns the distribution of cognitive work.
- Co-creation concerns the resulting interaction.
- Accountability concerns responsibility for its outcomes.
These constructs should not be treated as interchangeable. Together, however, they provide a more complete framework for understanding human–AI knowledge production.
AI-Aware Education Is Different From AI-Integrated Education
Recent discussions in higher education increasingly suggest that institutions must move beyond simply integrating AI tools into existing pedagogical practices.
AI-integrated education asks:
How can we incorporate AI into this course?
AI-aware education asks a more fundamental question:
How should this course operate now that generative AI exists?
The distinction is consequential.
An AI-aware assignment does not necessarily require students to use AI. An instructor may intentionally prohibit AI for a particular activity precisely because the intellectual effort required by that activity constitutes the learning objective.
For example, if students must learn to formulate an argument independently, delegating the initial construction of that argument to generative AI may undermine the intended learning process.
Later in the same course, however, students might appropriately use AI to generate counterarguments, simulate peer review, identify weaknesses, compare theoretical interpretations, or interrogate assumptions.
The pedagogical question therefore shifts from:
Can students use AI?
toward:
Which intellectual operations can be delegated without undermining the intended learning outcome?
Epistemic Delegation and the Intellectual Effort Inflection Point
Generative AI can reduce unnecessary cognitive and procedural burdens. It can accelerate formatting, facilitate brainstorming, generate alternatives, translate text, support preliminary exploration, or help users reorganize information.
These efficiencies can allow learners and researchers to devote greater attention to higher-order intellectual activity.
But there is a threshold.
Once AI begins performing the cognitive activity that constitutes the purpose of a learning experience, assistance becomes epistemic delegation.
Epistemic delegation occurs when a person transfers part of the cognitive work involved in producing, evaluating, interpreting, or communicating knowledge to another human or artificial system.
The important question is therefore not whether delegation occurs. Intellectual work has always depended on distributed systems of expertise, including collaborators, librarians, databases, calculators, software, and institutional infrastructure.
The critical question is whether learners and researchers understand what is being delegated, why it is being delegated, and what intellectual responsibility remains with them.
This is the problem addressed by the intellectual effort inflection point: the point at which productive assistance begins to replace intellectual effort that is itself consequential to learning, reasoning, or knowledge production.
From Individual Cognition to Communicative Orchestration
Human–AI knowledge production rarely involves only one human and one artificial system.
A student may interact with instructors, classmates, scholarly literature, databases, search engines, institutional policies, generative AI systems, software platforms, and multimodal communication environments during the production of a single academic artifact.
This creates a broader competence that extends beyond AI literacy.
Communicative orchestration is the capacity to intentionally coordinate human participants, artificial systems, information resources, communication modalities, and forms of expertise toward a situated objective.
Under this model, AI becomes one component within a larger communicative and epistemic ecology.
The student is therefore not merely an AI user.
The student increasingly becomes an orchestrator of distributed intellectual resources.
Responsible AI Co-Creation Requires Traceability
If human–AI interaction becomes part of knowledge production, simple disclosure may no longer be sufficient.
Writing “I used ChatGPT” reveals very little about the intellectual process.
Did the system correct grammar? Generate the research question? Suggest theoretical relationships? Search for sources? Summarize literature? Produce code? Interpret results? Challenge an argument? Draft the conclusion?
These activities represent substantially different forms of epistemic delegation.
A more accountable model requires distinguishing among three related practices:
Citation
Citation identifies an external source, system, or intellectual contribution when citation is appropriate.
Disclosure
Disclosure explains that artificial intelligence was used and describes the role it played in the intellectual process.
Documentation
Documentation preserves sufficient evidence of the human–AI process to permit evaluation, auditing, verification, or reconstruction when necessary.
Together, citation, disclosure, and documentation shift AI governance away from policing the mere presence of AI and toward understanding its function within intellectual production.
The Human Contribution May Become More Important as AI Improves
Increasingly capable generative AI may paradoxically increase the importance of human intellectual agency.
When producing plausible text, images, summaries, and alternatives becomes inexpensive, generation itself becomes less distinctive.
What becomes increasingly valuable is the ability to determine:
- What question deserves investigation?
- Which perspective is missing?
- Which generated proposition is defensible?
- Which evidence should be trusted?
- What should be rejected?
- When should AI be used?
- When should AI not be used?
- What does the resulting knowledge mean within a particular disciplinary, cultural, organizational, or human context?
These are not peripheral activities surrounding knowledge production.
They increasingly constitute its defining activities.
Toward AI-Aware Co-Creation
Generative AI requires higher education to reconsider more than its technology policies. It requires reconsideration of intellectual labor, assessment, collaboration, authorship, accountability, and the relationship between learning and production.
The central educational problem is therefore not reducible to AI prohibition versus AI adoption.
A more consequential question is:
What kind of human agency should education cultivate in a world where cognition, communication, and creation can increasingly be distributed across humans and artificial systems?
The answer should not be human independence from AI, nor human dependence upon AI.
Instead, I propose AI-aware co-creation: a form of asymmetrical human–AI collaboration in which generative systems contribute possibilities to an iterative intellectual process while humans retain epistemic agency, evaluative authority, situated judgment, and accountability for the resulting knowledge.
The future of education may therefore depend less on teaching students simply how to use AI and more on teaching them what to delegate, what to retain, what to question, what to verify, and what responsibility means when intellectual creation is no longer performed alone.
Frequently Asked Questions About AI-Aware Co-Creation
What is AI-aware co-creation?
AI-aware co-creation is an iterative human–AI knowledge-production process in which generative AI contributes possibilities while humans retain responsibility for purpose, evaluation, judgment, and the resulting intellectual product.
How is AI co-creation different from AI assistance?
AI assistance supports a predominantly human-directed task. Co-creation occurs when iterative interaction with an AI system influences the direction, formulation, development, or interpretation of the resulting intellectual object.
Does AI co-creation mean humans and AI are equal collaborators?
No. Human–AI co-creation is asymmetrical. Artificial systems do not possess equivalent intentionality, lived experience, ethical responsibility, situated judgment, or accountability. Co-creation describes the interactional process, not equality between participants.
What is epistemic delegation?
Epistemic delegation occurs when a person transfers part of the cognitive work involved in producing, evaluating, interpreting, or communicating knowledge to another human or artificial system.
What is epistemic agency in human–AI collaboration?
Epistemic agency is the capacity to determine what intellectual work should be performed personally, collaboratively, or delegated to artificial systems while retaining responsibility for evaluating the resulting knowledge claims.
What is the difference between AI-aware education and AI-integrated education?
AI-integrated education focuses on incorporating artificial intelligence into teaching and learning activities. AI-aware education begins from the broader recognition that generative AI exists and asks how learning outcomes, assessments, intellectual effort, and instructional practices should consequently be designed, including circumstances in which AI use should intentionally be limited.
Why is accountability important in human–AI co-creation?
Accountability makes the distribution of intellectual labor visible. Citation, disclosure, and documentation can help distinguish human contributions from delegated processes and provide evidence of how AI participated in producing an intellectual outcome.
References
Kornbluth, S. A. (2026). AI and education: A watershed moment at MIT. Massachusetts Institute of Technology.
León, C., Lipuma, J. M., & Oviedo-Torres, Y. (2025). AI in STEM education: A transdisciplinary framework for assessment, evaluation, and personalized learning. Frontiers in Education, 10, 1619888.
León, C. (2026). Intellectual effort inflection point and generative AI-driven epistemic delegation. Journal of Higher Education Theory and Practice, 26(2), 118–130.