AI is part of the workflow
Students can use generative AI for explanation, planning, code generation, debugging, refinement, exploration, and other legitimate intellectual work.
The educational implementation of Intellectual Agency
The AI-Integrated
Classroom.
The AI-Integrated Classroom integrates generative AI into learning while making the student's intellectual process visible, assessable, and accountable. AI can do more of the work; the student still owns understanding, judgment, verification, and the final result.
Core shift
The outcome is no longer the authoritative artifact. The process becomes part of the graded product. Read more on Medium
Traditional assumption
Final artifact
Learning inferred
AI-integrated model
Learning evidenced
The problem
The final artifact can no longer carry the weight we gave it.
Generative AI can produce convincing code, essays, explanations, plans, and solutions. A polished submission therefore tells us less than it once did about what a student understood, decided, verified, or actually contributed.
The AI-Integrated Classroom does not try to preserve the pre-AI classroom by banning AI, treating AI use itself as misconduct, or making detection the center of assessment. It changes what counts as evidence of learning.
The model
Integrate the AI.
Preserve the student's intellectual agency.
Students may use AI extensively. The requirement is not artificial isolation from the tool; it is continued human ownership of the intellectual work that matters.
Process becomes visible
Evidence is distributed across the work rather than inferred from one finished artifact. Development, decisions, documentation, and reflection matter.
Verification is required
AI producing an answer is not the same as the student establishing that it is correct. Testing, checking, and validation are explicit intellectual acts.
Responsibility stays human
The student remains responsible for understanding the problem, judging the output, correcting errors, explaining decisions, and owning the final result.
Evidence of learning
Assess the trace, not just the endpoint.
In the AI-Integrated Classroom, learning is evidenced across multiple artifacts and behaviors. No single item is expected to prove everything.
The final submission still matters. It is simply no longer treated as the sole authoritative record of the student's intellectual participation.
01Decomposition & development
How the work is broken down and how it evolves over time.
02Testing & verification
How correctness is established rather than merely assumed.
03Documentation
Whether the student can communicate what was built, how it works, and how it is used.
04Reflection & explanation
How AI was used, what was learned, what was questioned, and how output was evaluated.
05Final artifact
The outcome remains evidence—but one part of a larger evidentiary record.
One implementation in practice
Programming makes the model unusually visible.
In current programming courses, the framework is implemented through a development workflow that leaves a durable record of intellectual participation.
01AIUse the tool
02DevelopWork incrementally
03TestVerify behavior
04CommitPreserve history
05DocumentExplain the work
06ReflectAccount for AI use
Git & GitHub history
Students develop incrementally, test changes, make coherent commits, and preserve a history showing how the solution evolved.
Automated feedback
Automated tests provide continuous external feedback during development. They are primarily formative rather than a substitute for judgment.
README.md
Professional documentation requires students to describe the program, explain execution and usage, and communicate the software as a coherent artifact.
AI-use reflection
Students document how AI was used, what it helped them understand, how they challenged it, and how they verified what it produced.
What it is not
Not an anti-AI classroom.
Not an AI-surveillance classroom.
AI bansAI integration
Detection-centered enforcementVisible process evidence
AI use as misconductAI use with responsibility
Final artifact as proofDistributed evidence of learning
The limit of any instructional system
You cannot force someone to learn.
AI makes superficial completion easier in some ways—but it also makes deep learning more accessible than ever. The purpose of the AI-Integrated Classroom is not to build a perfect anti-cheating system. It is to make meaningful learning highly available, reward intellectual participation, make superficial completion less sufficient, and return responsibility for learning to the student.
The broader framework
Built on Intellectual Agency.
The AI-Integrated Classroom is the educational implementation of Intellectual Agency: a broader framework for using artificial intelligence as intellectual leverage while preserving human ownership of understanding, judgment, verification, and final responsibility.
Explore Intellectual AgencyProvenance
Created and implemented in practice.
The AI-Integrated Classroom was created by Alexander Katrompas, PhD, through the redesign of computer science and programming courses for an environment in which generative AI is a normal part of intellectual work. The model grows from classroom practice: integrate the tool, expose the process, require verification, and keep responsibility with the student.