Growth Mindset Emojifier Multimodal App

Yulia Kumar, Anjana Manikandan, Patricia Morreale, J. Jenny LiProceedings of the ... International Florida Artificial Intelligence Research Society Conference 2024 · DOI

TL;DR

This paper introduces the Growth Mindset Emojifier Multimodal App (GMEMA), an AI-powered tool that combines Growth Mindset principles with emojis to deliver personalized and emotionally engaging feedback for students. By leveraging OpenAI's models, GMEMA aims to enhance student engagement and learning outcomes, particularly in programming education.

What Problem Does This Paper Address?

Traditional educational feedback, while important, can sometimes be dry, unengaging, or fail to foster a growth mindset. Students may struggle to interpret critical feedback constructively, leading to decreased motivation. The authors address this by exploring how to make feedback more impactful and encouraging. They hypothesize that incorporating elements of the "Growth Mindset" framework, which emphasizes effort and learning from challenges, alongside the emotionally resonant medium of emojis, can significantly improve student engagement and comprehension. This is particularly relevant in technical fields like computer science, where complex concepts and challenging assignments are common.

Methods

The core of GMEMA is built upon OpenAI's advanced AI models, specifically ChatGPT-4-Vision and DALL-E 3. The application is designed as a dynamic web app using a Next.js front-end and Node.js platform. GMEMA has four main components:

  • GM (Growth Mindset): Utilizes ChatGPT-4 to enrich educational feedback with Growth Mindset principles and appropriate emojis.
  • Emojifier: This component is responsible for transforming standard feedback into a growth-mindset-oriented message, augmented with a relevant emoji. The paper provides a pseudocode for this process, outlining how it takes user messages, processes them with OpenAI for GM transformation, and appends emojis based on sentiment analysis.
  • Auto-Quiz Generator: This module automatically creates custom Java-based programming quizzes that are aligned with specific course content.
  • Auto-Grader: This component automatically evaluates student responses to the generated Java quizzes, providing immediate grading and feedback that is informed by the student's performance and the GM principles.

The system is designed to handle various input and output formats, leveraging AI for personalized feedback generation.

Datasets

The paper does not explicitly list specific datasets used for training the AI models. However, it references existing research on:

  • Growth Mindset literature (Dweck, 2009; Yeager, 2019).
  • Studies on emojis in sentiment analysis (Devlin et al., 2018; Eisner, 2016; Steel, 2020; Kim, 2022).
  • Research on Large Language Models and Transformer architectures for GM-focused content (Liu, 2019; Raffel, 2020).

The study also mentions a preliminary survey of over 150 students who showed a positive response to emoji-enhanced instructor feedback, which laid the foundation for this work.

Key Insights and Ideas

  • The Power of Emotional Nuance in Feedback: The study strongly suggests that incorporating emojis into educational feedback significantly enhances students' emotional engagement and comprehension. This highlights the potential of "emotional AI" in educational practices.
  • Bridging Growth Mindset and Emojis: GMEMA effectively merges the pedagogical framework of Growth Mindset with the engaging nature of emojis, creating a more holistic and motivating feedback loop.
  • Automated, Personalized Learning Tools: The integration of automated quiz generation and grading, coupled with AI-driven feedback, offers a scalable solution for personalized education, especially in technical subjects.
  • AI as a Catalyst for Engagement: The use of advanced AI models like ChatGPT-4 and DALL-E 3 demonstrates their capability to go beyond simple text generation, providing nuanced and contextually relevant feedback.

Key Results

The paper presents qualitative results in the form of example "Emojified Feedback" generated by the model for various quiz scores (ranging from 57% to 100%). These examples showcase how the feedback is tailored to the score, offering encouragement, suggesting areas for improvement, and maintaining a positive, growth-oriented tone, often accompanied by relevant emojis. For instance:

  • A score of 57% receives feedback like "Good start with 57% in your Java quiz! Keep practicing and focus on improvement. You're on your way! 👍"
  • A perfect score of 100% is met with "Outstanding with 100% in your Java quiz! Keep up the amazing work and explore new challenges. Superb achievement! 🚀"

The preliminary observations indicate a significant positive impact on student engagement and comprehension due to emoji integration.

Limitations and Future Work

Limitations:

  • The paper acknowledges that GMEMA is currently in its prototype stage.
  • The effectiveness is based on preliminary observations and a survey, suggesting a need for more extensive empirical validation.

Future Work:

  • Improving the app's capabilities, including further enhancing emojified feedback and quiz generation/grading modules.
  • Introducing text-to-speech and speech-to-text modalities by integrating OpenAI's Whisper AI model to enhance interactivity and inclusivity.
  • Expanding the application to support more programming languages and educational contexts.

Summary updated

Abstract

The study introduces the Growth Mindset Emojifier Multi-modal App (GMEMA), a novel application designed to improve educational feedback by including Growth Mindset and emojis. Using OpenAI’s AI models, GMEMA provides emotionally nuanced feedback to boost student engagement and foster growth mindsets. The application's capabilities extend to the automated creation and assessment of Java quizzes, delivering personalized feedback that blends motivational phrases and chosen emojis to match students’ achievements. Preliminary observations indicate that incorporating emojis into educators' feedback significantly heightens students' emotional engagement and comprehension, illustrating the profound impact of emotional AI on educational practices.
BibTeX
@inproceedings{kumar2024growth,
  author = {Yulia Kumar and Anjana Manikandan and Patricia Morreale and J. Jenny Li},
  title = {{Growth Mindset Emojifier Multimodal App}},
  booktitle = {Proceedings of the ... International Florida Artificial Intelligence Research Society Conference},
  year = {2024},
  doi = {10.32473/flairs.37.1.135572},
}
Figure