Feedback Review Guide

Table of contents


Overview

Feedback review is a critical quality control step in FeedForward. As an instructor, you review AI-generated feedback before it's released to students, ensuring it aligns with your teaching goals and maintains appropriate standards. This guide covers the review workflow, editing options, and best practices.

Understanding Feedback Review

Why Review AI Feedback?

  1. Quality Assurance - Ensure feedback meets your standards
  2. Personalization - Add instructor insights and context
  3. Safety Check - Catch any inappropriate content
  4. Learning Alignment - Verify feedback supports objectives
  5. Student Support - Identify students needing extra help

The Review Workflow

1. Student submits draft
   
2. AI generates feedback
   
3. Instructor notified
   
4. Instructor reviews
   
5. Instructor approves/edits
   
6. Student receives feedback

Accessing Feedback for Review

Review Dashboard

Your central hub for pending reviews:

Feedback Review Dashboard
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Pending Reviews: 12
├── ENGL101 - Essay 1: 8 submissions
├── ENGL101 - Essay 2: 3 submissions
└── ENGL201 - Research Paper: 1 submission

Recent Activity:
 3 new submissions in last hour
 15 reviews completed today
 2 flagged for attention

Quick Actions:
[Review All] [Filter] [Export] [Settings]
  1. From Dashboard - Click notification badge - Select "Pending Reviews" - View all awaiting feedback

  2. From Course - Navigate to course - Click "Assignments" - See review count per assignment

  3. From Assignment - Open specific assignment - Click "Review Feedback" - See all submissions

The Review Interface

Submission Overview

When reviewing a submission:

┌─────────────────────────────────────────────────┐
│ Student: Jane Smith                             │
│ Assignment: Essay 1 - Personal Narrative        │
│ Draft: 2 of 3                                   │
│ Submitted: 2 hours ago                         │
│ Word Count: 847                                 │
│ Processing Time: 45 seconds                     │
│                                                 │
│ AI Models Used:                                 │
│ • GPT-4: 3 runs (averaged)                     │
│ • Claude-3: 3 runs (averaged)                  │
│                                                 │
│ Overall Score: 82% (B)                          │
└─────────────────────────────────────────────────┘

Three-Panel Layout

The review interface shows:

┌─────────────┬──────────────┬─────────────────┐
   Student         AI          Instructor   
 Submission     Feedback        Actions     
                                            
 [Original    [Generated    [Edit Options]  
  Text]        Feedback]    [Scoring]       
                            [Comments]      
                            [Approve/Reject]
└─────────────┴──────────────┴─────────────────┘

Reading the Submission

Features for easier review:

  • Highlighting - Mark important sections
  • Zoom - Adjust text size
  • Search - Find specific content
  • Navigation - Jump between sections
  • Previous Drafts - Compare improvements

Understanding AI Feedback

Feedback Components

AI feedback typically includes:

Overall Assessment:
  Summary: Brief overview of submission quality
  Score: Numerical/letter grade
  Key Strengths: Top 2-3 positive aspects
  Main Areas for Improvement: Top 2-3 concerns

Rubric-Based Evaluation:
  For each criterion:
    - Score (points/percentage)
    - Specific feedback
    - Examples from text
    - Suggestions for improvement

Detailed Comments:
  - Paragraph-level observations
  - Writing mechanics notes
  - Content-specific feedback
  - Next steps recommendations

AI Confidence Indicators

Understanding AI certainty:

Confidence Levels:
🟢 High (90-100%): AI very certain about evaluation
🟡 Medium (70-89%): Some uncertainty in scoring
🔴 Low (<70%): Significant uncertainty, careful review needed

Common Low-Confidence Triggers:
- Unusual formatting
- Mixed languages
- Creative/unconventional approaches
- Technical jargon
- Very short/long submissions

Review Actions

Quick Approve

For high-quality AI feedback:

  1. Read through all feedback
  2. Verify accuracy and appropriateness
  3. Check scoring alignment
  4. Click "Approve & Release"
  5. Feedback sent to student immediately

Edit Before Approval

When modifications needed:

  1. Click "Edit Feedback"
  2. Editing Options: ``` Text Editing: - Modify any feedback text - Add personal comments - Remove inappropriate content - Adjust tone or language

Score Adjustments: - Change overall score - Modify rubric scores - Add score justification - Override AI calculations ```

  1. Save and Approve - Preview changes - Save edited version - Approve for release

Adding Instructor Comments

Personalize feedback:

Comment Options:
  Location:
    - Top of feedback (most visible)
    - After specific rubric items
    - Bottom summary

  Types:
    - Encouragement
    - Specific guidance
    - Resource links
    - Meeting requests

  Formatting:
    - Rich text editor
    - Bullet points
    - Links
    - Emphasis (bold/italic)

Example comment:

Instructor Note: Jane, I'm impressed with your improved 
thesis statement! For your final draft, consider adding 
one more supporting example in paragraph 3. Feel free to 
visit my office hours if you'd like to discuss further.

Requesting Regeneration

When AI feedback is inadequate:

  1. Click "Regenerate Feedback"
  2. Options: Regeneration Settings: - Use Different Model: [Select] - Adjust Temperature: [Slider] - Add Context: [Text field] - Focus Areas: [Checkboxes]

  3. Additional Context Example: "Student is ESL learner - provide more grammar explanations and examples. Focus on article usage and verb tenses."

Flagging for Attention

Mark submissions needing follow-up:

Flag Types:
  Academic Concern:
    - Potential plagiarism
    - Off-topic submission
    - Significant regression

  Student Support:
    - Mental health indicators
    - Request for help
    - Technical issues

  Technical Issue:
    - Corrupted file
    - Wrong assignment
    - System error

Bulk Review Features

Batch Processing

Handle multiple submissions efficiently:

  1. Select Multiple - Check boxes next to submissions - Or "Select All Similar"

  2. Bulk Actions: Available Actions: ✓ Approve all selected ✓ Add same comment to all ✓ Apply score adjustment ✓ Flag for later review ✓ Export for offline review

Quick Review Mode

For experienced reviewers:

Quick Review Settings:
├── Show only essential info
├── Keyboard shortcuts enabled
├── Auto-advance after action
└── Batch similar submissions

Keyboard Shortcuts:
A - Approve
E - Edit
R - Regenerate
F - Flag - Next submission - Previous submission

Review Filters

Organize your workflow:

Filter Options:
  By Status:
    - Unreviewed
    - In progress
    - Flagged
    - Approved

  By Score:
    - High (85%+)
    - Medium (70-84%)
    - Low (<70%)
    - Outliers

  By Student:
    - First-time submitters
    - Multiple drafts
    - Struggling students
    - High performers

  By Time:
    - Oldest first
    - Newest first
    - Due soon
    - Overdue

Review Best Practices

Consistency

Maintain fair standards:

  1. Review Similar Together - Group by topic - Compare approaches - Ensure fair scoring

  2. Use Rubric as Guide - Reference criteria - Apply uniformly - Document exceptions

  3. Track Patterns - Note common issues - Identify trends - Adjust instruction

Efficiency Tips

Streamline your process:

  1. Set Review Schedule ``` Suggested Schedule: - Morning: Complex submissions - Afternoon: Quick approvals - Evening: Batch processing

Time Allocation: - Quick approve: 1-2 minutes - Minor edits: 3-5 minutes - Major revision: 10+ minutes ```

  1. Use Templates - Common comments - Frequent corrections - Standard encouragements

  2. Prioritize Reviews - Final drafts first - Struggling students - Time-sensitive items

Educational Value

Maximize learning impact:

  1. Balance Criticism - Start with positives - Specific improvements - End with encouragement

  2. Be Specific ❌ "Improve your thesis" ✅ "Your thesis 'Technology affects society' is too broad. Try focusing on one specific technology and its impact, such as 'Social media has fundamentally changed how teenagers form friendships.'"

  3. Guide Next Steps - Clear action items - Resources to consult - Office hour invitations

Handling Special Cases

Problematic Submissions

When content concerns arise:

  1. Off-Topic Work - Don't approve AI feedback - Add instructor comment - Request resubmission - Note in gradebook

  2. Potential Plagiarism - Flag for investigation - Don't release feedback - Follow institutional policy - Document concerns

  3. Personal Disclosures - Handle sensitively - Consider privacy - Offer support resources - Consult if needed

Technical Issues

Common problems and solutions:

Garbled AI Feedback - Regenerate with different model - Check submission format - Manual feedback if needed

Missing Rubric Scores - Manually add scores - Check rubric configuration - Report to admin

Processing Errors - Resubmit for processing - Try different file format - Contact support

Analytics and Insights

Review Statistics

Track your review patterns:

Your Review Analytics:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

This Week:
Reviews Completed: 47
Average Review Time: 3.5 minutes
Edits Made: 23%
Regenerations: 5%

This Semester:
Total Reviews: 342
Quick Approvals: 65%
Major Edits: 15%
Regenerations: 8%
Flags: 12%

Efficiency Score: 87/100

Pattern Recognition

AI identifies trends:

Detected Patterns:
  Common Issues:
    - Thesis statements (45% of edits)
    - Citation format (30% of edits)
    - Conclusion weakness (25% of edits)

  Student Patterns:
    - 5 students consistently low scores
    - 3 students showing improvement
    - 2 students possible support needs

  Feedback Patterns:
    - GPT-4 better for creative writing
    - Claude-3 better for research papers
    - Morning reviews 20% faster

Integration with Grading

Export Options

Move feedback to gradebook:

Export Formats:
  CSV:
    - Student name
    - Assignment
    - Score
    - Feedback summary
    - Review notes

  LMS Integration:
    - Direct grade sync
    - Feedback upload
    - Rubric mapping

  PDF Reports:
    - Individual student
    - Class summary
    - Progress tracking

Grade Calculation

How scores become grades:

Score Mapping:
93-100% → A
90-92%  → A-
87-89%  → B+
83-86%  → B
80-82%  → B-
[continues...]

Custom Mappings:
- Curve adjustments
- Weighted categories
- Drop lowest
- Bonus points

Next Steps


Develop a consistent review routine - students appreciate timely, thoughtful feedback more than perfect feedback delivered late.
Remember that you're training AI to better understand your standards. Your edits and regeneration requests help improve future feedback quality.