Developing an AI Model to Identify Math & Literacy Instruction in Early Childhood Education Classrooms. Technical White Paper
Saved in:
| Title: | Developing an AI Model to Identify Math & Literacy Instruction in Early Childhood Education Classrooms. Technical White Paper |
|---|---|
| Language: | English |
| Authors: | Aravind Sundaresan, Leigh Ann DeLyser, Sarah Gerard, Gullnar Sy, Nancy Perez, John Niekrasz, Claire Christensen, SRI International |
| Source: | SRI International. 2025. |
| Availability: | SRI International. 333 Ravenswood Avenue, Menlo Park, CA 94025. Tel: 650-859-2000; e-mail: customer.service@sri.com; Web site: https://www.sri.com/ |
| Peer Reviewed: | N |
| Page Count: | 24 |
| Publication Date: | 2025 |
| Sponsoring Agency: | Gates Foundation |
| Document Type: | Reports - Research |
| Education Level: | Early Childhood Education Elementary Education Kindergarten Primary Education Preschool Education |
| Descriptors: | Video Technology, Artificial Intelligence, Technology Uses in Education, Mathematics Instruction, Identification, Automation, Kindergarten, Preschool Education, Literacy Education |
| Abstract: | A high-quality early childhood classroom provides children with a safe and nurturing environment to develop their physical, social, emotional, and academic capabilities. Children may receive literacy instruction during a morning circle story read-aloud, and then later break into small-group instruction focused on comparing quantities of dinosaur toys. Additional instructional moments also arise during informal play and learning, such as children counting off in line to go out to recess. Many early childhood providers use video to support teacher development, classroom observations, or for parents to check in on their children during the day. Researchers from SRI, supported by the Gates Foundation, explored opportunities to leverage AI and video from early childhood classrooms to explore AI-supported approaches to identify when math and literacy instruction happens. The research team used previously developed models that label academic content in YouTube videos, repurposing the models to measure their ability to detect instruction content in classroom settings. This technical white paper describes the research team's approach to automatically identifying academic content, the data, initial model performance, and implications of this work. |
| Abstractor: | As Provided |
| Entry Date: | 2026 |
| Accession Number: | ED679864 |
| Database: | ERIC |
Be the first to leave a comment!
