Abstract:The universal availability of ChatGPT and other similar tools since late 2022 has prompted tremendous public excitement and experimental effort about the potential of large language models (LLMs) to improve learning experience and outcomes, especially for learners from disadvantaged backgrounds. However, little research has systematically examined the real-world impacts of LLM availability on educational equity beyond theoretical projections and controlled studies of innovative LLM applications. To depict trends of post-LLM inequalities, we analyze 1,140,328 academic writing submissions from 16,791 college students across 2,391 courses between 2021 and 2024 at a public, minority-serving institution in the US. We find that students' overall writing quality gradually increased following the availability of LLMs and that the writing quality gaps between linguistically advantaged and disadvantaged students became increasingly narrower. However, this equitizing effect was more concentrated on students with higher socioeconomic status. These findings shed light on the digital divides in the era of LLMs and raise questions about the equity benefits of LLMs in early stages and highlight the need for researchers and practitioners on developing responsible practices to improve educational equity through LLMs.

What problem does this paper attempt to address?

This paper attempts to explore the impact of the practical application of large - scale language models (LLMs) such as ChatGPT in education on educational equity. Specifically, the study focuses on the following core issues: 1. **The impact of large - scale language models on the quality of academic writing**: The paper analyzes how the availability of large - scale language models affects the quality of students' academic writing. The study finds that with the popularization of large - scale language models, the overall writing quality of students has improved. 2. **Changes in the writing gap between students from different backgrounds**: The study specifically focuses on whether the writing quality gap between students with language advantages and disadvantages has been narrowed due to the introduction of large - scale language models. The results show that the writing quality of students with language disadvantages has improved more significantly, which helps to narrow the gap with students with language advantages. 3. **The impact of socioeconomic status (SES) on students' writing improvement**: The study further explores whether students with higher socioeconomic status benefit more from large - scale language models than those with lower socioeconomic status. The results indicate that although the writing quality of students with language disadvantages has generally improved, this improvement is more significant among students with higher socioeconomic status, which may exacerbate socioeconomic inequality. ### Research methods - **Data sources**: The study uses data from 1,140,328 forum assignments submitted by 16,791 undergraduate students in 2,391 courses at a four - year public university in the United States from 2021 to 2024. - **Measurement of writing quality**: Writing quality is measured by three comprehensive indices: readability, lexical diversity, and syntactic complexity. Each index is obtained by averaging the standardized scores of multiple standard linguistic indicators. - **Regression analysis**: A fixed - effects linear regression model is used to estimate the changes in writing quality before and after the introduction of large - scale language models, and the differences between students from different backgrounds are analyzed through interaction terms. ### Main findings - **Overall improvement in writing quality**: After the introduction of large - scale language models, the overall writing quality of students has gradually improved. - **Students with language disadvantages benefit more**: The improvement in the writing quality of students with language disadvantages is greater than that of students with language advantages, which helps to narrow the gap between them. - **The impact of socioeconomic status**: Although students with language disadvantages benefit, this improvement is more obvious among students with higher socioeconomic status, which may lead to the exacerbation of socioeconomic inequality. ### Discussion The study reveals the double - edged - sword effect of large - scale language models in education. Although these tools are helpful in improving the writing quality of students with language disadvantages, their socioeconomic impacts need further attention. The study emphasizes that when promoting and applying large - scale language models, a responsible approach needs to be adopted to ensure the fairness and inclusiveness of the technology. ### Conclusion The application of large - scale language models in education has potential, but their effects are not evenly distributed. In order to achieve educational equity, researchers, educators, and policymakers need to work together to develop and implement more responsible technology application strategies.

Whose ChatGPT? Unveiling Real-World Educational Inequalities Introduced by Large Language Models

ChatGPT and large language models in academia: opportunities and challenges

Embracing AI in Education: Understanding the Surge in Large Language Model Use by Secondary Students

Adapting Large Language Models for Education: Foundational Capabilities, Potentials, and Challenges

The GPT Surprise: Offering Large Language Model Chat in a Massive Coding Class Reduced Engagement but Increased Adopters Exam Performances

The rise of large language models: challenges for Critical Discourse Studies

Is ChatGPT Massively Used by Students Nowadays? A Survey on the Use of Large Language Models such as ChatGPT in Educational Settings

ChatGPT "contamination": estimating the prevalence of LLMs in the scholarly literature

"The teachers are confused as well": A Multiple-Stakeholder Ethics Discussion on Large Language Models in Computing Education

Fairness of ChatGPT

Gender, Age, and Technology Education Influence the Adoption and Appropriation of LLMs

LLMs as Research Tools: A Large Scale Survey of Researchers' Usage and Perceptions

Insights from Social Shaping Theory: The Appropriation of Large Language Models in an Undergraduate Programming Course

The Unequal Opportunities of Large Language Models: Revealing Demographic Bias through Job Recommendations

Unveiling the Competitive Dynamics: A Comparative Evaluation of American and Chinese LLMs

When Young Scholars Cooperate with LLMs in Academic Tasks: The Influence of Individual Differences and Task Complexities

LLeMpower: Understanding Disparities in the Control and Access of Large Language Models

"With Great Power Comes Great Responsibility!": Student and Instructor Perspectives on the influence of LLMs on Undergraduate Engineering Education

Unraveling Downstream Gender Bias from Large Language Models: A Study on AI Educational Writing Assistance

Practical and Ethical Challenges of Large Language Models in Education: A Systematic Scoping Review