India, July 29: Children have unprecedented access to digital video content, but identifying content that supports academic learning remains a challenge. A study by the Indian School of Business finds that the popularity of children’s videos does not necessarily translate into learning outcomes, highlighting a critical gap between what children watch and what they actually learn.
Introducing a novel AI-driven approach to assess the academic quality of children’s video content at scale, the research titled “Quantifying the Academic Quality of Children’s Videos Using Machine Comprehension,” authored by Professor Sumeet Kumar, Assistant Professor of Information Systems at the Indian School of Business, along with Mallikarjuna Tupakula and Ashiqur R. KhudaBukhsh of the Rochester Institute of Technology was published in the INFORMS Journal on Computing , and conducted in collaboration with the Rochester Institute of Technology.
The study analysed nearly 49,000 videos from 134 popular children’s YouTube channels drawn from a global pool of content, supported by a dataset of over 6,300 curriculum-based question–answer pairs and benchmarked against approximately 21,000 textbook questions spanning grades 1 to 12. The findings show that videos with higher viewership are not always those that deliver stronger educational value, pointing to a broader challenge in how digital platforms surface and prioritise content.
To address this, the researchers developed a machine comprehension-based model that evaluates videos using school curricula. By assessing whether a video can correctly answer textbook-aligned questions—using both its spoken content and visual elements—the model provides an objective way to measure its academic relevance. The approach achieved an accuracy of up to 77.5%, demonstrating its potential as a scalable evaluation tool.
“Parents today face a digital deluge of content, making it difficult to distinguish between what is educational and what is simply engaging,” said Sumeet Kumar, Assistant Professor of Information Systems at ISB. “Our research provides a systematic and scalable way to evaluate children’s videos using academic benchmarks. By aligning content with school curricula, we aim to help parents, educators, and platforms prioritise videos that genuinely support learning.”
The study also identifies gaps in content availability. While foundational topics such as basic measurement and language skills are relatively well represented, more complex areas—including scientific reasoning and advanced concepts—remain underrepresented. This points to the need for stronger alignment between digital content and curriculum requirements, particularly for higher-grade learners.
By providing a data-driven method to assess academic quality, the research offers practical implications for multiple stakeholders. For parents and educators, it enables more informed content choices. For digital platforms, it presents an opportunity to improve recommendation systems by incorporating learning value alongside engagement metrics.
As video continues to play a growing role in children’s learning environments, the study underscores the importance of moving beyond popularity as a measure of value. Strengthening the focus on educational quality will be critical to ensuring that digital content meaningfully contributes to learning outcomes at scale.