Transforming open and distance learning with generative AI: Custom micro-credentials from existing curriculums

Abstract

Short courses emerged under the name of e-learning in the 1980s and have become widespread in the digital environment worldwide with the advent of Internet accessibility. Applications that provide short and flexible learning, such as micro-credentials, are expanding globally, with higher education institutions like Open University offering this service internally. However, some institutions with large open and distance learning systems, such as Anadolu University, have yet to integrate micro-credentials. The research utilised simulation research and two-step agent-based modelling. In the first step, 100 micro-credential proposals were requested from the books of 522 undergraduate courses taught in the Anadolu University Open Education System through the generative artificial intelligence application designed over ChatGPT-4, MyGPT. This study explores the development of self-micro-credentials by adapting frameworks established by the Open University to Anadolu University. A different MyGPT was created in the second step, and it was requested that the units suggested under various names in the first step be exactly matched. The hallucination rate, initially 79.7% in Step 1, was significantly reduced to 5.3% through a structured validation process. The false negative rate was measured at 6.8%, indicating challenges in recognising domain-specific units. At the end of the two-step process, the accuracy rate was calculated as 92.3%, demonstrating the potential of generative artificial intelligence to facilitate self-micro-credential development. According to the findings, expert human oversight remains essential to ensuring reliable AI-generated recommendations, enabling learners to create tailored micro-credentials from distance learning resources aligned with their interests.