Give feedback on a batch of academic written assignments in higher education

Context

A university lecturer or instructor who needs to give written feedback on a batch of 15 or more student assignments — essays, lab reports, practicum reports, case analyses — from the same course. Rather than drafting each comment from scratch, the model generates a base comment differentiated by performance profile, which the instructor then adapts and personalises before delivery.

Paste before

Before opening the model, prepare the following:

• The rubric or assessment criteria you used, or a list of criteria with their weightings. If your rubric is in PDF or image format, transcribe the criteria as plain text — models without vision cannot process attachments.
• A summary of the patterns you observed when reviewing the batch: common errors, group strengths, recurring conceptual difficulties.
• Between three and five performance profiles that capture how you can group the assignments (for example: "conceptually strong but lacking argumentative structure"; "well written but no bibliographic support"; "incomplete or off-task").
• For each profile: an anonymous 4–6 line description of what those assignments have in common, without naming anyone.

Privacy rule — student data protection: Do not paste names, student ID numbers, enrolment codes, individual grades, or any data that could identify a student. In Chile, this is governed by Ley 19.628 and its 2024 update; in your context, apply your applicable data protection law (e.g., FERPA in the US, GDPR in Europe). The profile system exists to generate useful feedback without exposing personal information to an external service. If an assignment has characteristics that would identify a single student, describe them generically or handle it separately outside the model.

Act as a university lecturer in {{course subject}} with experience in giving formative feedback at scale in higher education. I will share the context of a course and the performance profiles I identified in a batch of written assignments. I need you to generate a base comment for each profile that I can adapt before delivering it to each student.

**Institution and course:** {{name of university}} — {{course subject}}
**Course level:** {{year or level in the programme, e.g. "second year of teacher education" or "fifth semester of engineering"}}
**Assignment type being assessed:** {{describe the genre: argumentative essay, practicum report, case analysis, critical review, etc.}}
**Expected assignment length:** {{e.g. "2,000 words" or "5 pages in APA format"}}

**Rubric or assessment criteria:**
{{paste your criteria or rubric here, with weightings if available}}

**General patterns across the batch:**
{{briefly describe what you observed when reviewing the whole set: common errors, group strengths, recurring conceptual problems}}

**Performance profiles (no identifiable data):**

*Profile A — {{descriptive name, e.g. "strong argumentation, insufficient sources"}}:*
{{describe in 4–6 lines what these assignments have in common, without naming anyone}}

*Profile B — {{descriptive name}}:*
{{description}}

*Profile C — {{descriptive name}}:*
{{description}}

{{add more profiles as needed — three to five is typical}}

**Instructions for each comment:**

0. Head each comment in bold with the format: **Profile [letter] — "[descriptive name]":** using the descriptive name I provided for that profile.
1. Write each comment as a continuous paragraph of 8–10 sentences, NOT as a bullet-point list. The comment should read as a written note from a lecturer to a student.
2. Internal paragraph structure: open by acknowledging the profile's clearest strength; identify the priority area for improvement precisely and directly; close with a concrete, achievable suggestion for the next assessment task.
3. Use the register standard in your institution — formal or informal second person, as appropriate for your course — and specify below.
4. Leave a {{here}} marker where I need to insert an observation specific to the individual assignment.
5. Tone: direct, respectful, and formative — neither condescending nor excessively complimentary.

**Grammatical register for the comments:** {{formal / informal second person, as used in your institution}}

Expected output

Example (abridged). Profile A is shown in full; B and C are trimmed — the model produces approximately 9 sentences per profile in all cases.

Profile A — "strong research question, shallow analysis": Your assignment demonstrates a real strength in constructing the research question: it is well scoped, clearly justified, and gives the rest of the report a coherent direction. {{here}} The methodological design is also well handled: the choice of qualitative approach is grounded, and the interview guide is consistent with your stated objectives. However, the analysis section is where the work loses depth: the interview excerpts appear in the same order they arose in the conversation, without being grouped into thematic categories or brought into dialogue with the conceptual framework you developed in your literature review. This means the analysis reads more as an annotated transcript than as an interpretive exercise. The concepts from your theoretical framework appear in an earlier section but do not return when you analyse the data, producing a fracture between theory and empirical evidence that weakens the report's central argument. The reflexivity section, for its part, appears as a closing sentence and does not fulfil its methodological purpose: positioning yourself as a researcher means making explicit how your social location, assumptions, or relationship to the topic may have shaped both the production of data and its interpretation. {{here}} For your next assessment task, I suggest a concrete exercise: before drafting the analysis, build a two-column table — on one side, excerpts from the interviewee grouped by theme; on the other, the theoretical concept each excerpt speaks to — and use that table as the skeleton for your writing.

Profile B — "categories present but theory absent": [abridged] Your assignment shows genuine command of the basic procedure of thematic analysis: you identify categories, assign interview excerpts to them, and structure the section with a logic that goes beyond chronological transcription — a significant methodological step forward. {{here}} The central problem is that those categories remain suspended: they are grounded in the data but are never anchored to the authors or concepts you introduced in your literature review... [~6 more sentences — the model delivers the full paragraph]

Profile C — "off-task or incomplete": [abridged] The review of your report reveals difficulties that go beyond formal aspects and affect the fulfilment of the assignment's core brief. {{here}} The fieldwork component — conducting a semi-structured interview with a real participant under informed consent — is the axis around which the entire report is organised, and its absence or substitution with another type of material makes it impossible to assess the learning outcomes the course aims to develop... [~5 more sentences — the model delivers the full paragraph]

Watch out for

• Do not paste identifiable data. Names, student IDs, enrolment numbers, or individual grades must not enter the model under any circumstances — apply your applicable data protection regulations (in Chile: Ley 19.628 and its 2024 update; elsewhere: FERPA, GDPR, or equivalent). The profile describes a pattern, not a person.
• The base comment is not the final version. If you deliver it identically to several students, it loses its formative value. Use the {{here}} markers to personalise it before sending.
• Be alert with the lowest-performing profile. The model tends to soften critical feedback to avoid sounding harsh. If the Profile C or D comment does not name the core problem clearly, prompt: "Rewrite the Profile C comment to be more direct about the problem with [criterion X], without losing the respectful tone."
• Academic integrity concerns. If an assignment shows signs of AI-generated writing or plagiarism, do not include it in this batch. Feedback on academic integrity requires an institutional process, not an automated comment.

Suggested follow-up

If the comments came out too generic, specify the weak criterion: "Profile B includes assignments where the problem is not structure but the absence of the student's own argument — everything is quotations without synthesis. Rewrite the comment focusing on that problem and suggest a concrete synthesis exercise." If you also need group-level feedback to share in class or on the course forum, ask: "Write a 5–6 sentence paragraph for the whole group that acknowledges the batch's strengths and identifies the two most frequent areas for improvement, without referring to any individual."