A Practical Guide for Supporting Formative Assessment and Feedback Using Generative AI
Project Overview
The document explores the integration of generative AI, specifically large language models (LLMs), in educational contexts, particularly in formative assessments and curriculum design. It highlights how LLMs can enhance formative assessment by generating personalized feedback, adaptive questioning, and clarifying learning objectives, which can lead to increased student engagement and better understanding of learning goals. However, it also raises concerns regarding the quality and appropriateness of AI-generated feedback, stressing the importance of using LLMs in a pedagogically informed manner to complement traditional teaching methods. Additionally, the document discusses the broader applications of generative AI in automating various educational tasks, such as question generation and grading, while considering the ethical implications and challenges involved in their implementation. Overall, the findings suggest that while generative AI has the potential to significantly improve educational practices and outcomes, careful consideration is needed to ensure these tools are used effectively and ethically within the educational landscape.
Key Applications
LLM for generating educational content and feedback
Context: Classroom settings across various grade levels and subjects, including formative assessments, open-ended assignments, and discussions on learning objectives. Target audience includes teachers and students.
Implementation: Teachers and educators use LLMs like ChatGPT to generate student-friendly learning objectives, create multiple-choice questions for assessments, and provide personalized feedback on student work and open-ended responses. This involves prompting the AI to analyze content and align outputs with curriculum standards.
Outcomes: Improved clarity of learning goals, enhanced engagement and ownership of learning by students, time-saving in assessment creation, immediate feedback opportunities, and more personalized learning experiences.
Challenges: Variability in the quality of LLM outputs, ensuring alignment with educational standards, risk of generic feedback, and the need for teacher oversight to ensure accuracy and relevance.
Implementation Barriers
Systemic
Overwhelming workloads and low pay for teachers limit their capacity to adopt new teaching methods. Challenges in integrating AI tools into existing educational frameworks and systems.
Proposed Solutions: Utilizing LLMs to automate aspects of formative assessments could alleviate some of the workload. Investing in training for educators and ensuring compatibility with current technologies.
Cultural
Resistance from students unfamiliar with formative assessment practices can hinder adoption.
Proposed Solutions: Implementing gradual transitions to student-centered practices and integrating LLMs as supportive tools.
Technical
Inconsistent quality of LLM-generated outputs may lead to ineffective assessment practices. Challenges in integrating AI tools into existing educational frameworks and systems.
Proposed Solutions: Encouraging iterative refinement of prompts and outputs by educators to enhance quality. Investing in training for educators and ensuring compatibility with current technologies.
Ethical
Concerns regarding plagiarism and the integrity of student work when using AI tools.
Proposed Solutions: Implementing best practices for ethical AI use, such as educating students on proper attribution and responsible use of AI.
Pedagogical
Resistance from educators about the effectiveness and reliability of AI-generated content.
Proposed Solutions: Providing research and evidence of AI's benefits in education, alongside professional development opportunities.
Project Team
Sapolnach Prompiengchai
Researcher
Charith Narreddy
Researcher
Steve Joordens
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Sapolnach Prompiengchai, Charith Narreddy, Steve Joordens
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai