Who Gets to Be an AI Person?

What a data-driven classroom experiment reveals about confidence, identity and the AI workforce gap

Duration: 30 minutes

The 5-minute warm-up activity PDF is attached above. 

Who This Is For: This lesson is designed for elementary school teachers, instructional coaches, STEM curriculum designers, and equity-focused program directors who work in K-12 settings and face a recurring challenge: they know the AI workforce gap is real, but they are unsure how to address it before students have already formed fixed ideas about who belongs in tech. It also speaks directly to education researchers, school administrators and diversity officers at EdTech organizations who are asking whether short-term interventions move the needle on student identity and career aspiration. If you have ever wondered whether a single well-designed activity can shift a child's sense of possibility in STEM or AI fields, this lesson was built for you.

Real-World Applications

After-school programs and summer learning initiatives across the United States are under growing pressure to demonstrate measurable outcomes in AI readiness and equity. The research behind this lesson was conducted in exactly that context: a summer program called A Fresh Squeeze on Data used hands-on activities including running a candy stand simulation and training a machine learning model through Google's Teachable Machine to introduce data collection, data bias and AI categorization to students in Grades 3 through 5. Program evaluators, grant writers and curriculum developers can use the findings from this study to make more honest claims about what short interventions can and cannot achieve, and to design follow-up programming that actually tracks the outcomes that matter over time.

The Problem and Its Relevance

The data shows something important but inconvenient: a well-designed AI curriculum can make students more comfortable with the subject, but comfort is not the same as career interest, and researchers found no statistically significant link between program participation and students' inclination to pursue AI or computing as a future job. This distinction matters enormously for anyone who funds, designs or evaluates STEM enrichment programs, because it suggests that confidence-building activities and career pathway development are two separate problems that require two separate strategies.

The more disorienting finding is that gender made no measurable difference in outcomes among third-through-fifth graders, not because the gender gap in AI does not exist, but because students this young may not yet have absorbed the social messaging that tells them it does. Research cited in the paper suggests that gendered career views tend to develop during Grades 6 through 8, which means the window for intervention is earlier than most programs target and the consequences of missing it are harder to reverse.

What the Research Measured and Why It Matters

Self-efficacy and why it is not one thing

Self-efficacy is a person's belief in their own ability to succeed at a given task. In this study, researchers measured it across three distinct dimensions: interest in AI-related subjects, comfort with learning AI content and general attitude toward computing. The program produced a statistically significant improvement in only one of those three dimensions: student comfort, which increased with a p-value of 0.010. The other two dimensions showed no significant change. This finding matters because educators and program designers often treat self-efficacy as a single dial they can turn up, when in fact it has components that respond differently to different kinds of instruction.

Career interest and what it takes to shift it

Career interest measures how much a student sees themselves in a profession, not just whether they enjoy a subject. The study found that career interest scores improved only slightly and not significantly after the program. The researchers suggest several reasons: the program ran for only two class periods, which may be too short for students to connect an activity to a life path; students at ages 7 to 11 may not have developed the cognitive framing needed to translate subject enjoyment into occupational preference; and variables like mentorship, role models and longer program duration were absent. For practitioners, this is a direct argument for longitudinal program design over one-time experiences.

Social Cognitive Career Theory and the environment that shapes belief

The study uses Social Cognitive Career Theory as its framework. This theory holds that a person's behavior and beliefs are shaped by their environment as much as by their individual traits. Applied to AI education, it means that low participation rates among girls and underrepresented students are not primarily a talent problem. They reflect a pattern in which limited learning opportunities and non-inclusive environments reduce self-efficacy over time, which in turn discourages career pursuit. The theory predicts that changing the environment, not just the curriculum, is what produces durable change.

Constructivist pedagogy as a design choice

The A Fresh Squeeze on Data program used a constructivist approach: students built knowledge through doing rather than listening. Activities were designed at varying difficulty levels and placed throughout the program rather than concentrated in a single final project. Students used open-ended discussion questions and worked in small groups to test ideas and learn from each other. This structure gave students repeated low-stakes opportunities to interact with AI concepts and reduced the intimidation that comes from treating AI as advanced or exclusive content.

The Bottom Line

Making a child comfortable with AI is a real and measurable achievement, but it is a first step, not a finish line. Programs that stop at comfort and claim to have addressed the pipeline problem are overstating their impact, and the students most affected by that overstatement are the ones who needed more than one good day in a classroom.

The gender gap in AI does not appear in third grade because it has not been taught yet. If educators wait until high school to address it, they are not closing a gap but measuring one that has already closed in the wrong direction. The most consequential AI equity work happens before students decide who they are allowed to become.

Reflection Prompt for Instructors

What is one thing your current AI or STEM programming measures, and what is one thing it does not measure that probably matters more? Write it down before the next planning meeting.

#AILiteracy #GenderEquityInTech #AIEducation #DataLiteracy #K12AI