Exploring the Learning Styles of "Semi-Humans": A Comparative Study of Nine Large Language Models

This study aims to explore whether large language models (LLMs) possess distinct learning styles and their characteristics, as well as to examine the stability of these learning styles. By integrating Kolb’s learning style theory with the Honey & Mumford Learning Style Questionnaire (LSQ), we systematically experimented with nine representative LLMs and investigated three core questions: Do LLMs exhibit unique learning styles? If so, what are the characteristics of the learning styles of different LLMs? How stable are these learning styles? The findings indicate that LLMs indeed demonstrate significantly diverse learning style characteristics, which can be categorized into three groups based on their learning styles. The first group displays clear traits of reflectors, theorists, and pragmatists; the second group exhibits high flexibility and adaptability in their learning styles; and the third group shows notable theoretical analysis and deep reflective styles. The study reveals that most LLMs possess a considerable degree of stability in their learning styles. This discovery marks a preliminary exploration of the learning styles of the emerging "semi-human" LLMs, highlighting the potential for creating a digital twin society that mirrors human society through increasingly human-like "semi-humans"