AI and Early Childhood Education: A Need for Robust Safeguards

Based on a brief by ICS Policy Fellows – Carol Cheatham, PhD (University of North Carolina), Erin Hamel, PhD (Winthrop University), Veronica Parker, PhD (Clemson University), Keller Anne Ruble, MEd (BridgeCare), and Dolores Stegelin, PhD (Clemson University).

Research suggests that AI can be an effective educational tool for young children, allowing for a more personalized learning experience. Several studies indicate that AI can be especially helpful in special education settings and for children who are learning a new language. AI tools currently used in classrooms and early childhood programs include smart robots and smart toys that can help facilitate conversations and build literacy skills, AI tools that support curriculum development, and personal assistants such as Siri or Alexa which can engage children and enhance their relationships with teachers or peers. The use of AI can also improve skills of creative inquiry, and theory of mind skills – the ability to understand that others may have thoughts, beliefs, desires, and intentions that differ from one’s own – which is crucial for developing empathy, and navigating complex social situations. Additionally, conversational AI applications can foster interaction in children who may struggle in traditional social settings. AI can also support teaching quality, parent-child interactions, early childhood assessment, and administrative tasks.

The Potential for Harm

However, it is clear that there is great potential for AI to negatively impact young children. Key concerns include privacy and data security, digital dependency and social interaction, bias and accessibility gaps, and content quality and ethical issues.

Privacy and data security: AI systems can collect and analyze enormous amounts of data and can identify patterns and trends useful in educational and other settings, allowing educators and other professionals to develop protocols to meet students’ changing needs. However, personal data collected and analyzed about children (e.g., sensitive information about behavior and preferences) raises serious questions regarding privacy and data security, particularly when data are collected without parental consent or adequate protections.

Digital dependency and social interaction: The increasing use of AI-driven applications and devices can lead to excessive screen time, negatively affecting children’s physical health (e.g., eye strain, reduced physical activity, improper sleep patterns). For young children, reliance on AI for entertainment and learning can lead to digital dependency, harming their ability to participate in non-digital activities, which can exacerbate struggles to engage in offline activities or face-to-face interactions. While AI applications and machine learning tools can foster social learning abilities, over-dependence on AI technology for social interaction may impede children’s ability to develop social and emotional skills, like empathy and face-to-face communication.

Young children need hands-on, direct experiences with objects in their indoor and outdoor environments to build knowledge and make brain connections. AI can supplement their learning but cannot take the place of play, hands-on exploration, or in-person education provided by teachers. All areas of child development depend on physical exploration and manipulation of objects in the world around them. Seeing objects on a screen can reinforce, but cannot replace, this experience.

Bias and accessibility gaps: AI systems, especially those employing machine learning, require vast amounts of data to learn patterns and make predictions. Because AI systems are only as good as the data they are trained on, they can produce biases. If the data contains biases, it can lead to biased, unfair and/or discriminatory outcomes. For children, this could mean exposure to AI tools that inadvertently reinforce stereotypes or provide unequal opportunities. Also, inherent biases in AI educational tools may favor certain learning styles or cultural contexts, providing a disadvantage for children from diverse backgrounds and disproportionally affecting marginalized communities. Additionally, AI and machine algorithms designed for educational settings may unintentionally favor students from more privileged backgrounds, intensifying educational disparities. Children from low-income families or underserved communities might not have the same access to AI learning tools, which can exacerbate existing educational inequalities, creating a digital divide.

Content quality and ethical considerations: The quality of AI-generated content tailored for children can vary extensively. This may cause concerns regarding the appropriateness, accuracy and educational value of the information provided to children. If AI tools are not meticulously selected, carefully curated, or regulated, they can propagate misinformation or content that is not developmentally appropriate. In addition, ethical considerations must also be considered. The ethical implications of using AI with children involve questions regarding consent, the appropriate role of AI in childhood development, and the long-term impact of early exposure to AI technologies. Collaboration among teachers, school administrators, parents, as well as policymakers is essential to establishing effective and ethical AI policies designed to ensure that AI systems are developed and implemented in educational institutions responsibly.

Robust Safeguards Are Needed

Key policy recommendations for the development and implementation of AI in early childhood education settings include:

• Ensuring data privacy compliance
• Guaranteeing adherence to the Children’s Online Privacy Protection Act of 1998 (COPPA)
• Ensuring algorithmic transparency
• Providing guidance and resources for AI professional development for educators
• Ensuring AI serves as a supplement to, not a replacement of, play-based learning and in-person education from teachers.

Comprehensive policy oversight and regulatory frameworks are critical to ensure AI benefits young children and does not worsen disparities within and between communities. As AI tools are being rapidly developed, brought to market, and adopted, there is a clear need for defined policies and guidance to support the field. Additionally, as AI is used more frequently in educational settings, professional development for teachers on the use of AI becomes increasingly important. Teachers and administrators need up-to-date information and training to use it effectively. Training should prepare educators to appropriately and ethically leverage AI for personalized child learning, administrative efficiency, and improved communication.

Thoughtful approaches to the use of AI with young children is as important as vetting the tools themselves. AI tools that are used for young children’s learning, development, or health should be based on data models for their specific use cases and validated by experts to maximize their effectiveness. These experts should include developmental psychologists, early childhood educators, pediatricians, experts in family systems, disability, accessibility, and systemic bias. Also, policies for AI in early childhood should include clear guidance on equitable deployment of AI tools, especially in early learning settings, as the absence of such guidance may exacerbate the digital divide by over- or under-use.

There is a pressing need to study the impacts of AI on the development of young children from infancy through the primary grades. The American Academy of Pediatrics expresses caution about potential overuse of AI and the negative impacts on cognitive, social, and emotional development of the young child. Special care and consideration is advised in developing policies and regulatory frameworks for how children and early childhood professionals engage with AI.

*Read the full brief for sources, and additional information on the opportunities, challenges, and recommendations for use of AI with young children.