Listen and learn
Personalised podcasts creating using artificial intelligence (AI) could be a more enjoyable and effective way of learning than traditional textbooks, new research has found. Researchers collected information about students that included their main subject, age group, interests and preferred style of lecturing.
They used these details to tailor an AI-generated podcast script that was specific to each student based on a chapter from a textbook. These included personalised references, such as using examples that mentioned the learner’s favourite music.
An audio version of each script was then created and the users were given multiple choice questions after listening to them to test how well they had retained the information. They also filled out a questionnaire about the experience.
The study found that overall, the podcasts created with AI were judged to be more enjoyable than using textbooks. The personalised episodes saw a significant improvement in learning outcomes, although this was subject-specific.
The study said: “While personalisation did not significantly enhance the overall enjoyment of the learning experience, it improved comprehension and learning outcomes in subjects where the personalised content aligned well with students’ interests and prior experiences.
“These findings suggest that AI-driven personalisation can make educational content more relevant and effective, especially when tailored to the learner’s context. Educators and content creators should consider integrating AI-generated podcasts into their teaching strategies to enhance engagement and comprehension. However, the varied impact of personalisation across different subjects highlights the need for further investigation into how and when personalisation should be applied.”
The research was named the best paper at the recent CHI Conference on Human Factors in Yokohama, Japan, and was published in the official conference proceedings.
