Evaluating AI Feedback in Writing Instruction - Annotators



Study Overview

This research partnership with Quill and the Chan Zuckerberg Initiative (CZI) develops standardized evaluation methods to compare AI-generated feedback against feedback from experienced educators. We're building benchmark datasets and evaluation protocols to measure how well AI systems perform in providing writing instruction feedback.


Product Description

Quill is an AI-powered writing platform that provides real-time feedback to students on their writing. Students will be completing three activities where they read an article and then complete a series of three sentence-level writing prompts. In each prompt, students build a sentence using evidence from the text.


Participant Responsibilities/ Research Activities

As annotators, you will be asked to review and annotate student responses by writing 2–3 sentences of feedback on each. In this role, you’ll bring your expertise as an ELA teacher to help build a high-quality dataset that reflects authentic classroom feedback practices 

(~30 hours per teacher per evaluation cycle, $50/hr).

Annotators are NOT required to use Quill with their students or do work within school hours.


Assessment Required

None


Agreements Required

  • Individual Contractor Agreements


Ongoing Use/Cost

Quill.org is free for all students and teachers. Partners in the study will also receive free access to Quill Premium, which provides teachers with additional data reporting and professional development.


Benefits

  • Contribute to AI education research - Help shape the future of AI writing tools by participating in research that improves feedback quality

  • Access to open-source protocols - Schools and educators will have access to the evaluation methods and datasets developed through this partnership, enabling broader adoption of AI writing tools


Compensation

Teacher annotators: $50/hour (~30 hours per evaluation cycle). It is ideal if you can contribute to all 3 cycles but not required.


Timeline

October - December 2025
Recruitment
January 2026 Onboarding
February - March 2026
Cycle 1
April - June 2026
Cycle 2
July - September 2026
Cycle 3