Abstract:

Underrepresented minorities, such as indigenous populations in low-income countries, face significant disparities in higher education outcomes. Tackling these inequities is vital for promoting diversity, equity, and inclusion in both academic and social realms. Failure to address these issues can lead to higher dropout rates, limiting the potential of these individuals and hindering the progress of their communities and the world at large. However, there has been limited progress in applying learning analytics solutions to address educational equity challenges in emerging economies.

Our study introduces an innovative two-part approach to bridge this gap. First, we used Natural Language Processing (NLP) techniques to analyze the content of academic materials in undergraduate courses (exams, study cases, assignments and other written content), identifying instances of gender and racial biases across various educational resources. Then, building on these insights, we developed a GPT in ChatGPT to help a group of 300 educators create more inclusive content for the academic community.

Results show that the implementation of our model led to a 78% improvement in the perceived inclusiveness of educational materials and a 24% decrease in the time educators spend preparing class content, for courses that have 2 to 4 hours per week in a 16-week schedule. These results highlight the potential for analytical tools to advance educational equity and suggest promising directions for future research and application in emerging and low-income countries.

Keywords:

Inclusiveness, GPT, Generative AI, Higher education.