Beyond Screen Time: What Research Tells Us About AI in Early Childhood Education

As educators and parents, we often approach the intersection of technology and early childhood education (ECE) with caution. However, Artificial Intelligence (AI) is already reshaping this landscape. A recent scoping review by Su and Yang (2022) analyzed 17 studies from 1995 to 2021 to understand how AI is actually being used with children aged 3 to 8.

The findings suggest that when integrated thoughtfully, AI is not just a passive screen experience—it can be a tool for creativity, literacy, and inquiry. Here are the key takeaways from the research.

1. It’s More Than Just "Coding"

While many assume AI education is strictly about learning to code, the research paints a broader picture. Studies indicate that AI tools significantly improve children's understanding of AI concepts (like machine learning and robotics), but they also enhance broader developmental skills.

Researchers found positive impacts on:

• Creativity and Emotion Control: Engaging with AI tools helps foster creative thinking and emotional regulation.

• Collaborative Inquiry: AI-interfaced robotic toys can stimulate higher mental functions, encouraging children to work together to solve problems.

• Literacy: Interactive AI has been shown to support literacy skills alongside computational thinking.

2. Two Main Approaches to Integration

The review identifies two distinct ways AI is currently being integrated into early childhood settings:

The Intelligent Tutoring System (ITS)

These are adaptive systems designed to personalize the learning experience. By using data to evaluate a child's cognitive state, these systems can adjust learning paths in real-time to suit individual needs. However, the review notes a current lack of ITS specifically designed for the unique developmental needs of young children, presenting a significant opportunity for future development.

The Internet of Toys (IoT)

This approach involves networked smart toys and robotics, such as PopBots or Zhorai99. These physical, embodied tools are particularly effective for this age group because they support social interaction and "unplugged" conceptual learning10. For example, the conversational agent Zhorai helped children understand machine learning through dialogue and visualization.

3. "Teaching" the Machine

One of the most effective pedagogical strategies identified was allowing children to "teach" the AI. Instead of just consuming content, children used tools like Google’s Teachable Machine or the PopBots curriculum to train models.

• Knowledge-Based Systems: Children performed best in understanding knowledge-based systems, followed by supervised machine learning¹³.

• Agency: When children train custom machine learning models, they shift their perception of smart agents, viewing them as fallible but helpful tools rather than magical entities.

4. The Challenges: Equity and Assessment

While the potential is exciting, the review highlights critical gaps. First, there is a "digital divide." A study involving 102 children across four countries found that children from higher socioeconomic backgrounds had a better understanding of AI concepts, suggesting that access remains a barrier.

Furthermore, while we have proof-of-concept studies, we lack rigorously validated assessment tools to measure exactly what children are learning from these interactions¹⁶. Future curriculum development must focus on creating sound measurements for AI literacy in early years.

The Verdict

AI in early childhood is not about replacing teachers with robots. It is about using intelligent tools to scaffold inquiry, creativity, and collaboration. As we move forward, the focus must be on embodied, hands-on experiences—where children are not just watching the AI, but teaching it.

Reference

Su, J., & Yang, W. (2022). Artificial intelligence in early childhood education: A scoping review. Computers and Education: Artificial Intelligence, 3, 100049.