AI and Young Children: How Do We Optimize Benefits and Mitigate Risks?

Key Principles in Adapting Artificial Intelligence to Promote the Optimal Health, Development, and Well-Being of Infants and Toddlers
Paul H. Dworkin, MD
This blog post is part of our initiative on AI and Early Childhood. Read more blogs from the series, and learn more about our National Common Agenda for Artificial Intelligence and Early Childhood, here.
Much discussion, debate, and research is increasingly focused on the transformational impact of Artificial Intelligence (AI) on society. While the impact of AI on children and youth is being actively explored, such efforts typically address such key issues as the potential of AI to support academic learning, the perils of AI compounding the adverse impacts of increased screen time, how students may misrepresent AI products as their own to complete assignments, and the extent to which AI may inform teachers’ lesson plans.
The critical aspects of brain and human development during early childhood, the first three years of life, suggest the particular importance of advancing our knowledge of the benefits and challenges of AI on the optimal health, development, and wellbeing of young children. Such features as the disproportionately dynamic brain growth that occurs in the first three years of life, the unique plasticity of brain cells during infancy that enables recovery from insults such as disease or injury, and the critical role of early childhood experiences in the formation of critical connections (i.e., synapses) among brain cells demand that we carefully scrutinize the impact of any new technology on young children. Similarly, the impact of AI on such developmental phenomena as synchrony, the “give and take” and mutual reciprocity between caregivers and infants, secure attachment between infants and their caregivers, and the eventual promotion of young children’s autonomy and independence as a milestone of early childhood, demand our careful attention.
While the impact of AI on infants and very young children is mediated through its use by caregivers, how to best align the evolving technology with the realities of brain growth and infant development is uncertain. Can AI strengthen caregivers’ protective factors known to promote more optimal developmental outcomes, such as how to access concrete support at times of need, parents’ understanding of developmental themes and stages, and the strengthening of parenting skills? Alternatively, does caregiver reliance on AI restrict developing a warm and secure relationship with their infants, undermine their confidence in their intuitive tendencies, and fail to respect cultural beliefs and practices?
When confronted by new challenges, I am inclined to consider key principles that may come into play to inform our thoughts, strategies, and actions. Our work promoting children’s optimal health, development, and wellbeing over many decades has suggested a lengthy list of such critical concepts that are helpful in formulating approaches to novel, critical, contemporary issues. Several come to mind as potentially useful in considering how we may best harness both the utility of AI in promoting the optimal wellbeing of very young children and the risks posed to their development, particularly with respect to emerging social-emotional and cognitive skills.
I propose that we consider six of these key principles in the context of maximizing the benefits of AI for very young children and mitigating potential risks. Because research in this area is relatively rudimentary, contemplating such concepts is likely to raise more questions than suggest interventions or policies. My hope is that such considerations will stimulate thought and debate and, ultimately, encourage the important investigations to inform wise and impactful actions.
- Employ human-centered design as a key methodology to ensure that interventions are optimally aligned with target populations for desirability, feasibility, and viability. Human-centered design deeply engages the target populations for tools, processes, and interventions to ensure consideration of users’ needs, concerns, priorities, and preferences. In addition to engaging early childhood caregivers in the design and use of AI, examples of other key collaborators include product designers, child development and behavioral health specialists, and educators.
- Invest in technology and innovation that improve access for all members of society. The perils of the “digital divide,” i.e., the broadband or connectivity desert that so often characterizes underserved urban core and rural neighborhoods, are often cited as adversely impacting children’s learning and academic performance and as an important social determinant of health. Furthermore, generational differences in comfort with and uptake of modern technologies by the elderly may be particularly important when grandparents play a key role in the lives of young children. Limited access to the benefits of AI that impacts underserved populations is likely to further exacerbate such critical disparities.
- Provide culturally competent and sensitive information that acknowledges and respects the diverse backgrounds and beliefs of families and communities. Differences in deep-seated cultural beliefs and practices among families will impact views on such tools as AI. For example, Latino families traditionally value the concept of familism (familismo), which prioritizes the family as the key source of guidance and support for children’s development (Dumont-Mathieu T, Bernstein B, Dworkin P, Pachter L. Role of pediatric health care professionals in the provision of parenting advice: a qualitative study with mothers from four minority ethno-cultural groups. Pediatrics 2006;118:e839-848.) Among such families, AI may be regarded as competing with the role of parents and other key child caregivers in the nurturing of children’s wellbeing.
- Recognize the efficacy of strength-based approaches to enhance families’ protective factors (resilience, social connections, concrete support in times of need, understanding of children’s development and parenting skills, and capacity to advance children’s social–emotional development) to promote young children’s optimal health, development, and wellbeing. For example, AI may be particularly helpful in strengthening parenting skills, knowledge of child development, and ensuring families’ access to concrete support in times of need.
- Partner with the intended recipients of programs, services, and interventions to determine the appropriate measures, metrics, and outcomes that constitute efficacy and impact. The efficacy of various interventions is often determined by innovators who are seeking to impact specific, measurable outcomes considered important to return-on-investment, scaling, and sustainability. Such considerations may fail to take into account the priorities and concerns of target populations, e.g., parents of young children. Failure to consider the extent to which families regard outcomes for young children as important priorities will likely undermine the uptake and utility of interventions.
- Research the efficacy of innovations and prioritize dissemination of efficacious, evidence-based products. The dissemination of tools, processes, and interventions such as AI among caregivers to young children, including family and childcare providers, should be justified by research evidence of feasibility, desirability, and positive impact. Such evidence is critical to distinguish between short-term fads and sustainable best practices. Furthermore, implementation and dissemination must encompass quality improvement processes that ensure effective application.
Embracing these concepts does not guarantee that AI will be helpful, rather than harmful, in supporting families and communities in their efforts to promote the optimal wellbeing of infants and toddlers. Such pronouncements are dependent on demonstrating both the efficacy and safety of AI through well-designed research studies that yield generalizable and actionable results. Nonetheless, our experience does suggest that sound conceptual thinking can, and should, yield hypotheses that are worthy of careful investigation and findings that demonstrate both favorable impacts and unanticipated adverse outcomes. Our formidable challenge is to keep pace with the dynamic trajectory and extraordinary capacity of this remarkable technology, such that the desire for rapid dissemination does not overwhelm our capacity to invoke reason and sound judgement to inform AI’s trajectory for our youngest children.
Paul Dworkin, MD, recently served as the Executive Vice President for Community Child Health at Connecticut Children’s Office for Community Child Health and is the Founding Director of the Help Me Grow National Center. He is professor emeritus of pediatrics at the University of Connecticut School of Medicine. For 15 years, Dr. Dworkin previously served as physician-in-chief at Connecticut Children’s and chair of pediatrics at UCONN. He currently serves on the ICS Board of Directors.