The study employed a topic modeling approach to analyze a corpus of 4,175 articles related to generative AI in education. The findings suggest that while large language models (LLMs) like ChatGPT dominate the current educational research, other modalities such as text-to-speech, text-to-image, and text-to-video have received less attention, despite their potential to enhance personalized learning, problem-solving, and creativity.
The topic modeling analysis identified 38 research topics organized into 14 thematic areas, including Domains (e.g., geoscience, chemistry, business, and engineering education), Personalized Learning Support (e.g., text-to-speech, sentiment analysis, and feedback systems), Problem Solving (e.g., mathematical and physics problem-solving, simulations, and explanations), Technology Adoption, Professional Development, Creativity, Serious Games, Tools and Content (e.g., automated question generation, content design), Assessment, Ethics and Security, Integrity, Chatbots, and Language Learning.
The results highlight the need to extend research beyond text-to-text technologies and explore the broader potential of multimodal approaches in various educational domains and levels, including K-12 and higher education. The study also emphasizes the importance of addressing ethical and security concerns, academic integrity, and the impact of generative AI on teaching and learning practices.
To Another Language
from source content
arxiv.org
Principais Insights Extraídos De
by Vill... às arxiv.org 09-26-2024
https://arxiv.org/pdf/2409.16376.pdfPerguntas Mais Profundas