Units
1. Introduction to Artificial Intelligence in Pedagogy: Roles, Boundaries, Validation, and Student Data 2. Lesson Plan and Learning Objectives: Outcome, Bloom's Taxonomy, and Flow 3. Instructional Material Production: Presentation, Worksheet and Activity 4. Differentiated Instruction: Level, Readiness, and Inclusivity 5. Question and Quiz Production: Item Writing and Distractor Quality 6. Measurement-Evaluation and Rubric Design: Criteria, Level and Consistency 7. Feedback Generation: Formative, Constructive, and Personalized Feedback 8. Parent Communication: Message, Bulletin, Interview and Difficult Situations 9. Student Data Privacy and KVKK: Anonymization and Secure Use 10. Academic Integrity and Verification: Student Use of Artificial Intelligence 11. Ethics, Bias, End-to-End Workflow, and Teacher AI Governance
Unit 7 / 11

Feedback Generation: Formative, Constructive, and Personalized Feedback

Gains:

  • Ability to understand the principles of formative feedback and produce concrete, forward-looking and constructive feedback drafts with the support of artificial intelligence
  • Ability to write personalized feedback based on a rubric or set of criteria, indicating the student's strengths and developmental steps
  • Ability to understand that feedback should target the product, not the student, and that the artificial intelligence draft should not be given without verification by the teacher's observation.

The grade you give a student is just a number; What really moves him forward is feedback—messages that show him what his work is doing well and how he can improve. Research tells us that good feedback is one of the most powerful influences on learning. But writing personalized, constructive feedback to a class of 30 students takes hours; This is where teachers are most constrained. AI quickly generates feedback outlines based on the rubric and the teacher's observation; you personalize, validate and set the tone. Boundary is important: Feedback should be based on the student's actual work and the teacher's observation; Praise/criticism generated by AI "out of thin air" is not given without verification.

Principles of good feedback

Formative feedback is feedback given while learning is ongoing and helps the student progress; leaves room for correction, as opposed to the final note (summative). Principles of good feedback:

  1. It targets the product, not the person. It's not "you're lazy," it's "the concluding paragraph doesn't adequately support your idea." What is discussed is the student's work, not his identity.
  2. Concrete and specific. "It's nice" doesn't teach anything; "You gave three examples, this strengthens your claim" teaches.
  3. Forward-looking (feed-forward). It doesn't just tell you the error, it shows the next step: "add a question to the introduction and you'll grab the reader's attention."
  4. Balanced. One strength + one improvement step; Only criticism breaks motivation, only praise stops development.
  5. It's manageable. Not 10 mistakes in a study, but 1-2 most important development points; The student cannot correct all of them at once.
Tip: The "two stars, one wish" formula works: two concrete strengths (stars) and one improvement proposal (wish). If you give this formula to AI, it will produce balanced feedback.

The power and danger of AI

AI writes gentle, structured feedback paragraphs based on a rubric + teacher notes and personalizes it for 30 students. But there are two dangers. First, the AI ​​may produce generic, hollow praise (“great job”) without seeing the student's work—this is false feedback, and the student will notice. Second, the language of AI may be too uniform and mechanical; If 30 students receive the same sentence, the feedback becomes worthless. Solution: Give the AI ​​your actual observation for each student (what he/she did to what extent) and have it produce the draft accordingly, then add the human touch.

Step by step: Feedback generation with AI

  1. Give the criterion/rubric. What will the feedback be based on?
  2. Summarize your observation. Short, anonymous note such as "This student is strong on the content, weak on the conclusion."
  3. Specify the format. "Two stars, one wish", tone (supporting), length (3-4 sentences).
  4. Get the draft. Observational feedback for each student.
  5. Verify and personalize. Is the praise real, is the improvement step feasible; Add your own voice.
  6. Tone control. Is it age appropriate, non-discouraging, respectful?

three mini cases

Case 1 — Returning from empty praise. A teacher told the AI ​​to "write positive feedback" without giving it the work; The AI ​​produced general praise such as "a magnificent piece of work." The teacher realized it was fake; This time, he gave each student's rubric scores and an observation sentence. The feedback became concrete: "It is powerful to refer to 3 different sources when using sources; it would be clearer if you summarize the conclusion section in one sentence."

Case 2 — Time and personalization. A Turkish teacher spent ~4 hours per week writing feedback on 28 essays. He gave the rubric levels and one-line observation for each student to the AI ​​and had it produce a draft of “two stars, one wish,” then added a personal sentence to each one. The time went down to ~1.5 hours; students realized that the feedback was personal to them because it was based on real observation.

Case 3 — Tone correction. A teacher's AI feedback was correct but harsh: "your input is weak, your writing is unorganized." For a 6th grader, this was discouraging. The teacher said, "Rewrite in language that is age-appropriate, encouraging, and targets the product." The feedback became “you can make your introduction a little stronger; try starting with a question” — same message, constructive tone.

Copiable templates

1) Personal feedback based on rubric:

Your role: a supportive [subject] teacher. Below are a student's (anonymous) rubric levels and my one-sentence observation. Write a forward-looking feedback in the "Two stars, one wish" format, 3-4 sentences long, targeting the product (not the person). Age-appropriate, respectful tone.Rubric levels + observation: [paste]

2) Collective feedback draft (for the class):

Below are the criterion scores and an observation note of 5 students (Student A-E, anonymous). Write a different, personalized, non-repetitive feedback for each (two strengths + one improvement step). Don't let it feel like copy-paste. Data: [paste]

3) Softening hard feedback:

The feedback message below is correct, but its tone is harsh and does not encourage the student. Rewrite it in language that is constructive, targeted to the product, and shows the next step for a [X] year old student. Soften the message but maintain the context. Feedback: [paste]

4) Self-assessment question generation:

For "[task]", write 4 self-assessment questions that will make the student look back at their own work (e.g. "what did you improve the most, what would you change if you did it again"). Keep it age appropriate and thought-provoking.

Weak prompt / Strong prompt

Weak prompt:

Write good feedback to this student.

Without seeing the study and observation, AI produces generic, hollow praise; it becomes fake and useless.

Powerful prompt:

Your role: supportive Turkish teacher. Anonymous student, 6th grade composition. Rubric: content 4/4, organization 3/4, language 3/4, conclusion 2/4. My observation: the ideas are rich, but the concluding paragraph feels rushed. Write a 4-sentence, encouraging, product-targeted feedback in the "two stars, one wish" format.

poor feedback

strong feedback

Difference

"It's nice."

"Giving three examples strengthens your claim."

concreteness

"You're lazy."

“The conclusion does not support the idea.”

Product, not person

"You have many mistakes."

"Focus on this one thing first:..."

Manageability

"Do better."

“Try adding a question to the introduction.”

forward step

Common mistakes

  • Asking for feedback without giving the work. AI produces hollow, fake praise; the student notices and trust is lost.
  • Targeting the person. Phrases like "lazy/underachieving" hit self-confidence, not learning.
  • Just criticism or just praise. One breaks motivation, the other stops development; balance is necessary.
  • Uniform feedback. Same sentence to 30 students; Without personalization, feedback becomes worthless.
  • Too many corrections. Listing 10 mistakes in a study paralyzes the student; Give 1-2 priorities.
Attention: Feedback is also an output; Verify the factual claims in it (e.g. "sources are wrong") with your own observation. The AI ​​may misread the student's work and write an unfair critique.

In summary

Feedback is the bridge that turns grade into learning; The product targets, is concrete and forward-looking, is balanced and contains a manageable number of recommendations. AI quickly and personalizedly produces feedback drafts based on the rubric and the teacher's actual observation; This enables qualified feedback on a classroom scale. But praise produced without seeing the work is fake, uniform feedback is worthless, and the tone should be age appropriate. The observation is from you, the draft is from AI, the final say is from you.

Application task

Anonymize rubric levels and an observation sentence for the 3 student works you have (or hypothetically). Generate different feedback for all three with the "rubric-based personal feedback" template. Check each feedback with two questions: (1) is the praise based on actual work, (2) is the improvement step concrete and actionable. Then add a sentence from your own voice to each feedback.

checklist

  • [ ] Is the feedback based on the student's actual work and my observation?
  • [ ] Did I target the product and not the person?
  • [ ] Is there a strength and an actionable improvement step?
  • [ ] Is the feedback personalized or uniform?
  • [ ] Have I kept the tone age-appropriate and encouraging?
Previous Next