More than Model Documentation: Uncovering Teachers' Bespoke Information Needs for Informed Classroom Integration of ChatGPT

Title of the Paper

More than Model Documentation: Uncovering Teachers’ Bespoke Information Needs for Informed Classroom Integration of ChatGPT

Paper Information

• Subject Area: Applications of Generative AI in Educational Technology and Teacher Support
• Keywords: Large Language Models, ChatGPT, AI in Education, Machine Learning Documentation, Teacher Information Needs

Research Background and Problem

• Research Questions:

  • The integration of ChatGPT into classrooms bypasses typical training and validation procedures, leaving teachers to directly face the multifunctionality of generative AI.
  • Current transparency and documentation in educational technology are insufficient to meet teachers’ specific needs.
  • The lack of guidance tailored to teachers’ practices hinders the effective integration of ChatGPT into classrooms.

• Significance of the Research:

  • Teachers need to quickly evaluate ChatGPT to determine its suitability for classroom teaching objectives, which is critical for the rational application of educational technology.
  • The misuse of AI technology poses risks to teaching quality and the diverse learning needs of students.

• Motivation and Related Work:

  • Literature shows that teachers lack necessary support and training in technology integration, particularly in understanding emerging technologies like generative AI.
  • Current AI documentation frameworks are overly technical and fail to effectively assist teachers in integrating AI into learning scenarios.
  • This study aims to identify the information gaps teachers face and propose an interactive documentation framework to enhance technical support for teachers.

Solution

• Methods and Solutions:

  • The authors conducted interviews with 22 middle school English Language Arts (ELA) and social science teachers to explore their information needs and experimental behaviors.
  • A framework was proposed to help teachers dynamically understand the capabilities of generative AI and bridge the gap between technical and pedagogical knowledge.

• Innovations:

  • Connecting AI transparency and documentation practices to the integration of educational technology, the study proposes a new documentation framework for teachers to explore and apply generative AI.
  • Categorizing teachers’ information needs and contrasting them with the shortcomings of current documentation practices.
  • Designing an interactive documentation system to support teachers in testing model functionalities based on instructional goals.

• Implementation Steps and Techniques:

  • The research design includes two stages of interviews: open exploration and guided experimentation.
  • Identifying teachers’ behavioral patterns and challenges in information retrieval and experimentation during the interviews.
  • Proposing a structured five-step interactive documentation framework:
    • Define learning objectives
    • Analyze student needs
    • Select generative tasks
    • Provide prompt engineering support
    • Guide classroom coordination

Research Findings

• Specific Findings:

  • Identified significant information gaps and misconceptions among teachers, including misunderstandings of ChatGPT’s capabilities and a lack of scenario-relevant use cases.
  • Proposed an experimental documentation framework that connects technical functionalities with educational needs for teachers.
  • Generated content and guidance to help teachers explore AI applications that align with subject-specific requirements.

• Comparison with Existing Solutions and Advantages:

  • Current static documentation limits teachers’ discovery and active learning; the new framework supports teachers in dynamically understanding the technology through an interactive structure.
  • The framework introduces classroom role coordination and enriched application scenarios, addressing the abstraction of existing documentation.

• Experimental or Evaluation Results:

  • Interviews revealed that teachers prefer interactive technical support over text-based instructions.
  • Teachers struggled to envision complex application scenarios using generic documentation; the multi-layered pathways designed in the framework helped overcome this failure of imagination.
  • After iterative improvements, teachers were able to gradually handle the shortcomings of model outputs more effectively.

• Limitations and Future Directions:

  • The new framework cannot fully address all integration barriers and serves only as a supplementary tool to static documentation and ethical disclosures.
  • Future research could explore mechanisms for direct collaboration between AI developers and educators, as well as expand studies on teacher professional development.
  • Further work should focus on broader educational groups, emphasizing teaching equity and fostering students’ metacognitive skills.