Five steps, with room to reconsider
Start with an assignment you already teach. These five steps connect a focused response with the student’s next action and your judgment about what follows.
- Identify what most needs attention. Ask what is obstructing learning or its demonstration, why it matters now, and what you can reasonably change. GPF calls that obstacle an interference.
- Focus the feedback. Use a few targeted Green Pen marks or comments to identify a difficulty or a promising question. Give enough direction to begin and enough room to discover something you did not anticipate.
- Name the principle that travels. Name the idea guiding the inquiry or revision so the student can use it in another passage, task, or setting.
- Ask for an action you can examine. A “Show me” prompt asks the student to revise, explain, compare, or apply the idea. Choose an action that will help you judge what to do next.
- Interpret the response and decide what follows. Examine the change, consider other explanations, and decide whether to continue, adjust, stop, or raise a design or policy question.
A signal is an observation worth examining, whether it appears in the original work or in the student’s response. Its meaning remains open to interpretation.
The order provides a starting point. Steps can overlap, recur, or change order: a student may attempt a revision before articulating its principle. What holds the method together is the connection between the problem, the action, and the interpretation. New evidence can send the instructor back to the first step.
Three questions behind the method
What is the student discovering?
An unsettled claim may signal confusion or a discovery worth pursuing. Ask what disturbed the student’s first answer, then examine the evidence together. The teacher’s interpretation must remain open to revision too.
Can the teacher act on the evidence?
Responsibility requires authority. Faculty need time to examine learning, room to explain and reconsider a judgment, and a route to change assignments or policies that repeatedly obstruct it.
Can the student use the learning elsewhere?
If a college claims that learning carries beyond a course, it needs evidence beyond success on one task. A different context, purpose, or level of assistance can test how far the claim holds.
A focused teaching prompt
Illustrative teaching example: A paragraph cites a source but does not explain how it supports the claim. The instructor focuses on the missing connection and names the principle: evidence needs an explanation of its relevance.
Explain how this source supports your claim. Then find another place where the reader needs the same connection and revise it. Explain why you chose that passage.
The instructor now has specific choices to examine. A successful repair in the marked passage shows that the requested change appears. A sound explanation and a useful application elsewhere offer additional evidence of understanding. If the connection remains unclear, the next response may need to address source interpretation rather than sentence structure.
What GPF adds
Focused feedback and revision are established teaching practices. GPF connects them to an explicit account of judgment: why this response, what did the student’s action reveal, and who can address the problem that remains?
A colleague can examine and adapt that reasoning. A course leader can use it to consider a recurring assignment problem. The framework gives individual teaching decisions a place in professional conversation and institutional improvement.
Keep the work proportionate: one priority, an action within an existing assignment, and a brief record only when it helps the next decision.
Choosing the problem is part of the work
Sometimes the assignment is part of the problem. A rubric that asks for a firm position too soon may make a thoughtful question look like a weakness. A student who argues “it depends” may have a clear position if they explain what it depends on. Before correcting the student, check how you are reading the rubric. If the criteria rule out a well-reasoned response, the assignment may need revision.
Choose the obstacle that most blocks the intended learning. If several students omit the same requirement, examine both their understanding and the assignment’s design. The missing-component case follows that decision in practice.
What the evidence can tell us
The examples show the method in writing instruction; effectiveness across instructors and disciplines remains an open question.
A better draft alone does not show what a student can do independently. AI, a tutor, or a classmate may have helped with the revision and explanation. What the response tells us depends on the task and the help available. GPF cannot verify who did the work or replace observing a student perform the task.
If the work stays the same, ask why. The student may have misunderstood, had no opportunity to respond, or disagreed with the advice. Note what happened, consider possible explanations, and avoid claiming more than the evidence shows.
Ask another question only if the answer could change your next teaching decision. Stop when you have enough evidence for that decision. A short note about the problem, the requested action, the student’s response, and any remaining question may be enough. The work must also fit the time instructors have.
What GPF adds to an AI conversation
ChatGPT or Claude can suggest a useful question or help think through a teaching problem. GPF asks the educator to explain why a response fits, what evidence might change that judgment, and who is responsible for the next decision. Then the instructor follows what happens when the response is used.
Instructors remain responsible for the decisions they have authority to make. When a problem requires changes to a course or policy, they can bring the evidence to those who can act. Ochse’s scholarship calls this broader concern “evaluative sovereignty”: faculty need authority to exercise professional judgment and must be able to explain their decisions. GPF offers one way to put that idea into practice.
Try one response and examine what follows
Use the one-page Five Steps card or fillable response template with an existing assignment. Consider what the student’s response supports, what changes your interpretation, and what you would do next.
For the broader inquiry into faculty authority and evidence of learning, explore Roger’s scholarship. Related work includes Carless and Boud on student feedback literacy and TEQSA on assessment reform in the age of AI.
Principles for human judgment
Read the ten underlying commitments
Fluent performance is not sufficient evidence of learning.
Artificial intelligence can produce accomplished intellectual products. Fluency may demonstrate performance, but it does not by itself establish whose judgment produced the work or what human capability changed.
Learning is a change in human capability.
Education is not complete when an acceptable product appears. Learning requires growth in what a person can understand, judge, explain, revise, or do with increasing independence.
The submitted text is evidence, not proof.
A paper, solution, presentation, or project is one source of evidence. Green Pen Framework does not assume the reliability of the finished text; it asks for choices, revisions, and explanations that can be examined alongside the conditions of assistance. These observations inform judgment without guaranteeing independent learning.
Assessment connects visible action with careful interpretation.
Assessment should reveal meaningful decisions as work develops. A repair, an explanation, and a new application support different inferences. Faculty must consider what each observation establishes, what assistance was available, and what remains uncertain.
Faculty require evaluative sovereignty.
Faculty need recognized authority to decide what matters in a particular case, explain their reasons, and reconsider in light of evidence. Rubrics, metrics, templates, policies, and algorithms may organize evaluation; they cannot eliminate the need for accountable human discretion.
Student agency requires consequential intellectual action.
Students need room to discover a question or conclusion that the assignment did not anticipate. Rubrics should make standards for evidence and reasoning clear without prescribing the intellectual destination. Faculty must be able to recognize a promising departure, examine its grounds, and challenge criteria that penalize it.
AI capability does not confer educational authority.
A system may be capable of generating feedback, assigning a score, or recommending a decision. Capability alone cannot determine whether that delegation is educationally legitimate, who authorized it, or who remains answerable for the result.
Judgment does not scale.
Institutions can focus judgment, sustain it, and learn from its signals. They cannot reproduce situated human judgment simply by accelerating production, standardizing its appearance, or automating its language.
Private faculty labor cannot indefinitely repair public design failures.
When the same problem recurs, the evidence may point beyond individual teaching to assignment architecture, curricular sequence, assessment design, or policy. Sustainable practice requires institutions to address failures at the level where they are produced.
Human agency must be designed into educational systems.
Institutions preserve human agency by assigning meaningful authority, consequential action, and visible responsibility to faculty and students. Agency must be built into practice, curriculum, assessment, and policy—not merely affirmed in institutional language.