The central question driving the project is deceptively simple but increasingly difficult to answer: how do young children encounter, understand and make sense of artificial intelligence, and what does that mean for their development at a moment when AI tools are threading themselves into nearly every dimension of daily life? Silvis and her collaborators at the University at Buffalo argue that early childhood — the period when children form their most foundational understandings of how the world works — has been largely absent from the national conversation about AI in education. Most research, policy debates and classroom interventions have focused on older students, high schools and universities, leaving preschools, daycares and early elementary classrooms to navigate a fast-changing technological landscape without a map.

“As AI is increasingly integrated into our society, the project addresses the urgent need for equitable AI education in the early years, when children are forming foundational understandings of the world,” Silvis said. Her scholarship, which examines the relationship between technology, child development and social change, has long focused on how children interact with digital tools and how those interactions shape — and are shaped by — the social contexts around them. The new project extends that line of inquiry into territory she believes has been neglected: the years before children can even articulate what an algorithm is, when their encounters with AI are mediated by voice assistants, adaptive apps, smart toys and the adults in their lives.

The researchers identify two interconnected challenges that define the current landscape. The first is invisibility. AI education, to the extent it exists at all, remains largely absent from early childhood curricula, professional development programs and policy frameworks. Young children are already interacting with AI systems — asking questions of voice-activated devices, watching algorithmically curated videos, playing with toys that respond to their speech — but the systems that structure their learning rarely acknowledge this reality, let alone equip caregivers and teachers to respond to it. The second challenge is fragmentation. Early childhood education in the United States is delivered through a patchwork of settings: public pre-kindergarten programs, private preschools, childcare centers, family home-based care and informal learning environments. Each of these settings operates under different regulations, funding streams and professional cultures, which means that even when promising approaches to AI literacy emerge in one context, they rarely travel to others. Fragmentation, the researchers argue, compounds the invisibility problem, scattering responsibility for a systemic issue across dozens of disconnected institutions.

The Vision Grant is designed precisely to address this kind of complexity before a full-scale research project is launched. Over the next twelve months, Silvis and her colleagues at the University at Buffalo will build the project’s conceptual research framework while cultivating community partnerships across Western New York. This is not a phase of data collection in the conventional sense; rather, it is a deliberate period of groundwork intended to ensure that the eventual four-year research agenda is grounded in the lived realities of the people it aims to serve. The team will work with local education leaders, community members and technology partners to co-design a research plan focused on solutions that are not only effective in local settings but hold the potential to scale nationally — a criterion that is increasingly central to how the Spencer Foundation evaluates transformative education research.

The emphasis on community partnership reflects a broader shift in education research methodology, one that treats teachers, caregivers and families not as subjects of study but as co-producers of knowledge. For a topic as charged and fast-moving as AI, this approach carries particular weight. Parents of young children are often navigating AI tools with little guidance, forming their own intuitions about what is safe, appropriate or beneficial. Early childhood educators, meanwhile, face the dual pressure of preparing children for a world saturated with intelligent systems while lacking training, curricular materials or institutional support to do so. Technology companies, for their part, are marketing AI-enabled products to families and schools faster than research can evaluate them. A planning grant that brings all of these stakeholders to the same table is, in effect, an attempt to build the coordination infrastructure that the fragmented early childhood sector currently lacks.

The stakes of this work extend beyond classroom pedagogy into questions of equity, which the researchers place at the center of the project. AI literacy — the ability to understand, question and critically engage with intelligent systems — is rapidly becoming a form of foundational knowledge, much like numeracy or early literacy. When AI education arrives late in a child’s schooling, or not at all, the benefits accrue unevenly, often along lines of family income, community resources and access to well-resourced schools. Children who encounter AI only as consumers of its outputs, rather than as learners capable of understanding how it works and whose interests it serves, enter a world shaped by these technologies at a structural disadvantage. Silvis and her colleagues contend that waiting until middle or high school to introduce these ideas is too late; the understandings children build in their earliest years shape how they relate to technology for the rest of their lives.

Silvis joined SUNY Cortland in 2023 and holds a Ph.D. in learning sciences and human development from the University of Washington’s College of Education, one of the leading programs in the country for research on how people learn across settings and over the life course. The learning sciences tradition, which integrates insights from cognitive science, developmental psychology, sociology and design, is well suited to a problem like AI in early childhood, where the relevant questions span what children can comprehend, how technologies are designed, what families and teachers do in practice, and how policies and institutions shape all of the above. Her published work has appeared in a range of peer-reviewed outlets, including the International Journal of Computer-Supported Collaborative Learning, the Journal of Early Childhood Literacy, Information & Learning Sciences, the International Journal of Child-Computer Interaction, Learning, Culture and Social Interaction, and Cognition & Instruction — a portfolio that traces a consistent intellectual arc from children’s collaborative learning with technology to the social and cultural conditions that make such learning equitable.

The collaboration with the University at Buffalo also situates the project within a broader institutional ecosystem in Western New York, where university researchers, school districts and community organizations have increasingly partnered on education innovation. Building those relationships during the planning year is not incidental to the project’s goals; the researchers view the partnerships themselves as part of the eventual solution. A research agenda produced in isolation from practitioners, they argue, would replicate the very fragmentation the project seeks to remedy. Instead, the team hopes the planning process will surface the specific conditions, constraints and assets of local early childhood settings — knowledge that can then inform both the four-year research design and the practical tools it produces.

The Spencer Foundation’s Vision Grant program occupies a distinctive niche in the education research funding landscape. Rather than funding only mature, fully specified research projects, it invests in the upstream work of agenda-building: convening partners, testing conceptual framings, conducting pilot conversations and developing the theoretical scaffolding that ambitious long-term studies require. The Transformation Research Grant that Silvis’s team will ultimately pursue is the foundation’s flagship mechanism for supporting research aimed at rethinking education systems rather than incrementally improving them. By securing a Vision Grant, the team has effectively been invited to make the case that AI in early childhood education deserves that level of sustained, system-oriented attention.

Whether the field rises to meet that challenge remains an open question, but the timing of the project is notable. Artificial intelligence is already embedded in the media children consume, the devices their families use and the platforms their schools are beginning to adopt — often without any intentional educational design. Researchers in child-computer interaction and the learning sciences have documented how children anthropomorphize voice assistants, attribute knowledge and intentions to AI systems, and absorb tacit lessons about how information works from the technologies around them. What has been missing, Silvis and her colleagues argue, is a coordinated research agenda that treats these early encounters as a matter of educational equity rather than an afterthought of technological change. The next twelve months of planning in Western New York will determine whether that agenda — and the four-year study behind it — can become a model for how the nation prepares its youngest learners for an intelligent, automated world.

At its core, the study confronts the critical need for AI literacy at the elementary level—a challenge that becomes increasingly urgent as AI technologies permeate society at an accelerating pace. The researchers argue that early education must evolve beyond traditional boundaries to equip children not only with computational skills but also with the capacity to engage critically and creatively with AI. In this vein, the STEAM (Science, Technology, Engineering, Arts, and Mathematics) framework serves as an ideal platform to embed artificial intelligence into broader learning contexts, fostering interdisciplinary thinking and problem-solving.

One of the notable features of the course under scrutiny is its integration of Project-Based Learning (PBL), an instructional approach that encourages active exploration and real-world problem solving. By situating AI concepts within tangible projects, the curriculum stimulates student engagement and makes complex ideas more accessible. Moreover, the innovative inclusion of the Artificial Intelligence of Things (AIoT) component introduces children to the dynamic synergy between AI and IoT technologies, highlighting how data-driven intelligence manifests in everyday objects and environments.

The researchers employed a quasi-experimental methodology to rigorously evaluate the course’s impact, comparing student outcomes before and after program implementation while controlling for confounding variables. This design offers a robust lens to discern causal effects, especially in educational contexts where randomized control trials may be impractical or unethical. Additionally, the study’s emphasis on questionnaire validation ensures that the instruments measuring AI literacy and gender attitudes are both reliable and valid, thereby underpinning the credibility of their findings.

Results indicate a significant increase in AI literacy levels among students who participated in the STEAM–PBL–AIoT course. These gains encompass not only theoretical understanding but also practical skills in AI applications, algorithmic thinking, and ethical considerations. This multidimensional improvement underscores the efficacy of project-driven, interdisciplinary instruction in cultivating robust AI competencies in elementary learners, a critical step toward democratizing technology education from a young age.

Perhaps more striking is the study’s focus on gender equity, a persistent challenge in STEM education worldwide. By analyzing engagement and achievement metrics disaggregated by gender, the researchers were able to identify shifts in participation rates, self-efficacy, and interest levels between boys and girls. Encouragingly, the STEAM–PBL–AIoT curriculum contributed to narrowing the gender gap, fostering an inclusive classroom climate that values diversity and empowers all students to see themselves as capable AI practitioners.

This gender-sensitive approach is reinforced by curricular and pedagogical choices designed to counteract stereotypes and biases that often deter girls from pursuing STEM subjects. For instance, by incorporating collaborative projects and emphasizing creative problem-solving over rote memorization, the course creates an environment where diverse learning styles are accommodated and success is attainable for everyone. Such nuances in design may serve as a blueprint for wider educational reforms geared toward equitable AI literacy.

The integration of AIoT within the curriculum also serves as a salient element in bridging theoretical knowledge with tangible technological applications. AIoT exemplifies the convergence of intelligent algorithms with connected devices, a domain rapidly expanding in real-life settings such as smart homes, healthcare, and urban infrastructure. By introducing young learners to AIoT, the course resonates with contemporary technological trends and equips students with contemporary skill sets that transcend traditional disciplinary silos.

From a technical standpoint, the instructional design incorporates scalable AI tools tailored for beginner-friendly interaction. These include visual programming environments, interactive simulations, and sensor-based experimentation kits that enable hands-on experience. Such technologies demystify AI concepts, reducing cognitive barriers and allowing students to experiment with AI model training, data input, and decision-making processes. This tangible engagement is pivotal for solidifying abstract computational ideas.

Ethical literacy forms an integral component of the course, addressing the socio-technical implications of AI deployments. Given the profound societal shifts instigated by AI, educators must instill a sense of responsibility and critical awareness among learners. Discussions around AI bias, privacy, algorithmic transparency, and societal impact are embedded throughout learning modules, preparing students not just as technologists but as conscientious citizens capable of navigating the complex AI-powered world.

The researchers underscore the importance of rigorous questionnaire validation to ensure the accuracy of measuring AI literacy and gender equity outcomes. Developing and fine-tuning survey instruments that reflect students’ cognitive and affective dimensions of learning requires methodical psychometric analysis. Validation processes such as factor analysis, reliability testing, and pilot studies contribute to constructing assessment tools that generate meaningful and interpretable data.

Beyond immediate academic gains, the study’s implications are far-reaching. By establishing evidence-based strategies for fostering early AI literacy with a gender-equity lens, the research offers policymakers, curriculum developers, and educators practical insights to inform scaling efforts. In an era where technological proficiency is indispensable, creating inclusive entry points into AI education is vital for cultivating a diverse and empowered future workforce.

This work also serves as a call to action for more longitudinal studies tracking the sustained impact of AI education initiatives, especially concerning gender participation trajectories beyond elementary school. Understanding how early interventions influence long-term STEM engagement and career choices remains a crucial research frontier. Furthermore, adapting the STEAM–PBL–AIoT framework to varied sociocultural contexts offers promising avenues to enhance global AI literacy equity.

In summary, this pioneering study situates itself at the nexus of emerging educational needs and technological evolution. By methodically blending a comprehensive STEAM curriculum, immersive project-based learning, and cutting-edge AIoT integration, it charts a transformative path toward equitable AI literacy in formative educational stages. The results illuminate how thoughtfully designed educational interventions can dismantle gender barriers and build foundational AI competencies essential for tomorrow’s innovators.

As the world rapidly embraces AI-driven transformations, empowering all children to understand and harness AI technology is more than an educational imperative—it’s a societal one. This research exemplifies the profound potential of combining pedagogical innovation, technological toolkits, and equity-focused frameworks to cultivate a generation not just ready for the AI age, but poised to shape it responsibly and creatively.

With these foundational insights, educators and stakeholders are encouraged to reexamine existing curricula and pedagogies, ensuring inclusive access to AI education. The matrix of STEAM, PBL, and AIoT presents a compelling model that can inspire widespread curricular reforms and investment in teacher training, resources, and infrastructural support. Ultimately, this trajectory points towards a future where AI literacy and gender equity coalesce to generate richer scientific ecosystems and societal well-being.

Subject of Research: AI literacy development and gender equity in elementary education through STEAM–PBL–AIoT pedagogical interventions.

Article Title: AI literacy and gender equity in elementary education: A quasi-experimental study of a STEAM–PBL–AIoT course with questionnaire validation.

Article References:
Cheng, CC., Wang, JS., Zhai, X. et al. AI literacy and gender equity in elementary education: A quasi-experimental study of a STEAM–PBL–AIoT course with questionnaire validation. IJ STEM Ed 12, 50 (2025). https://doi.org/10.1186/s40594-025-00574-y

Image Credits: AI Generated