HYDERABAD: The popularity of children’s videos on YouTube Kids (YTK) does not necessarily reflect their academic value, according to a study by the Indian School of Business (ISB), which has developed an artificial intelligence-based tool to assess the quality of online video content.
The study, Quantifying the Academic Quality of Children’s Videos Using Machine Comprehension, analysed nearly 49,000 videos from 134 popular YTK channels. Researchers compared the videos against more than 6,300 curriculum-based question-answer pairs and about 21,000 textbook questions covering classes 1 to 12. They found that the most-viewed videos were often not the ones with the strongest academic outcomes.
Conducted by ISB Assistant Professor Sumeet Kumar along with Mallikarjuna Tupakula and Ashiqur R Khudabukhsh of the Rochester Institute of Technology, the study was published in the INFORMS Journal on Computing.
The researchers developed a machine comprehension-based AI model that evaluates both spoken and visual content to determine whether a video can answer curriculum-aligned textbook questions. The model achieved an accuracy of up to 77.5%.
Sumeet said the tool could help parents, educators and digital platforms identify content that genuinely supports learning. The study also found that while basic topics such as language skills and measurement are well represented online, advanced concepts and scientific reasoning remain underrepresented.