Sci-GiFD Platform

Science GenAi-Feedbacked Drawing in Supporting Students' Scientific Modelling

Background

Despite the promotion of GenAI in K-12 science education locally, there is a lack of subject-specific GenAI tools for facilitating high quality of science learning.

More importantly, developing interfaces with GenAI platform in science education require joint expertise from learning sciences and science education.

Scientific Models in Education

They are representations of mechanisms and interactions among entities to explain a phenomenon (Cheng and Min, 2015). In the topic of evaporation, students can draw how water particles evaporate from the surface of clothes (Hokayem and Schwarz, 2014), making sense of the scientific phenomenon.

Learning sciences principles

The platform, which uses multimodal GenAI, needs to ensure both scientific accuracy and observing principles of learning sciences: Cognitive interaction, artifact-mediated interaction and social-emotional interaction. It can provide domain-specific affordances in assessing and feedbacking to students’ scientific models.

Our Innovative Platform

This pilot project draws on synergistic research expertise in learning sciences from Seoul National University and that in science education from EduHK, to pilot GPT-4o-powered Sci-GiFD platform.

Prof Cho and Dr Cheung will develop it, exploiting the potential of GenAI in providing personalized feedback in improving students’ scientific modelling.science learning.

Project Components

Case Study Approach

Phase 1:
- One class of pre-service science teachers (n=20) will test and evaluate the Sci-GiFD model.
Phase 2:
- One class of seventh grade students (n=30) will engage in science instruction assisted by the Sci-GiFD platform.
- Students will use drawing and writing to represent their models of evaporation.
- They will upload their scientific models into the platform.

Sci-GiFD Assessment

Assessment of Students' Models in Sci-GiFD
- Models will be assessed based on:
  - Explanation
    - Attained: Students use the model to explain processes and mechanisms.
  - Evidence
    - Attained: Students mention how specific evidence can construct and change the model.
  - Prediction
    - Attained: Students map aspects of the model to other situations.
- Reference: Hokayem & Schwarz (2014)

Potential outcomes

The project will yield international co-authorship of one peer-reviewed conference paper and one journal article between two universities.

Demo

Our Team

Principal Project Supervisor(s) (PPS)

Dr Cheung Ka Ching, Kason

Assistant Professor of STEM Education

Co-Project Supervisor(s) (Co-PS)

Prof Young Hoan Cho

Professor, Director of the Learning Sciences Research Institute