Extending Computer Science to AI Literacy: The Case for Data Science Education

Abstract

As artificial intelligence (AI) becomes increasingly embedded in schools, workplaces, and everyday digital systems, K-12 computer science education faces a growing question: how should students be taught about AI and prepared to engage with it? AI literacy frameworks emphasize competencies beyond programming and algorithmic thinking, including understanding how AI systems operate, evaluating their outputs, and recognizing their limitations and ethical implications. This article develops a conceptual argument that data science education offers an important pathway for preparing students for AI because many contemporary AI systems are grounded in data. Drawing on literature in AI literacy, computer science education, and data science education, as well as our research on high school data science curricula and implementation in Utah and Virginia, this article examines how data science is emerging across courses, curricula, and state-level initiatives. The analysis highlights variation in instructional emphasis, course placement, student access, and teacher preparation needs. It argues that data science can extend computer science education by creating interdisciplinary, data-centered opportunities for students to encounter computing concepts in applied contexts and develop the foundational understanding needed to engage critically and responsibly with AI.

1. Introduction

As AI becomes more embedded in digital tools, schools, and workplaces, students need preparation for a world in which AI will shape both daily life and work. In response to this shift, Wang and Lester (2023) call for new K-12 AI literacy curricula. These calls, however, require clarity about what AI literacy entails and what students should be expected to learn. AI literacy has been defined as a broad set of competencies, including understanding how AI works, using and evaluating AI, creating with AI, and attending to AI’s ethical dimensions (Long & Magerko, 2020; Ng et al., 2021). Preparing students for an AI-driven world therefore requires more than technical familiarity alone.

These developments raise important questions for computer science education and its contributions to AI literacy. Current AI literacy scholarship suggests that programming and algorithmic thinking, while important, are not sufficient on their own for helping students understand AI (Long & Magerko, 2020; Ng et al., 2021). It also points to the importance of helping students understand how AI systems operate, what their limitations are, and the broader social and ethical issues they raise (Long & Magerko, 2020; Ng et al., 2021; Zhou et al., 2025). Emerging concerns that routine reliance on AI tools may reduce the critical engagement and independent reasoning that AI literacy depends on (Risko & Gilbert, 2016; Tian & Zhang, 2025) make this especially pressing.

This AI literacy scholarship increasingly acknowledges that AI systems are built on data, implying that students need to understand how data are gathered and used—yet it rarely addresses how or where in K-12 education that understanding should develop. This paper addresses that gap. We argue that the data practices underlying AI—how data are collected, labeled, curated, and interpreted—are central to AI literacy, and that students can learn these practices through the growing field of data science education. In doing so, we treat AI literacy not as the province of computer science alone, but as something built across disciplines wherever students work seriously with data. Drawing on the AI literacy, computer science education, and data science education literature, as well as our research on high school data science, we show how data science education can extend computer science education toward AI literacy.

2. AI Literacy, Computer Science Education, and Data Science

2.1. AI in K-12 Education

Literature on AI in K-12 settings points to two broad ways AI has entered classrooms. One is through the tools and systems that students and teachers use (Ghosh & Kumar, 2026; Miao et al., 2021). AI is present in student-facing applications such as intelligent tutoring systems, adaptive learning platforms, and writing and feedback tools, as well as in teacher-facing tools used for assessment, administration, or monitoring student progress (Miao et al., 2021). Understanding how AI is embedded in these tools may help students recognize its presence in their learning environments, but such exposure alone does not help them understand AI’s capabilities and limitations, critically evaluate AI outputs, or use AI effectively and ethically (Zhou et al., 2025).

A second way AI may enter K-12 education is through instructional materials that treat AI more directly as subject matter content within computer science, mathematics, and related subjects. Touretzky et al. (2019), for example, argue that core AI ideas should be introduced into the K-12 curriculum, and Wang and Johnson (2019) demonstrate meaningful connections between AI concepts and topics already taught in high school mathematics. Michaeli et al. (2023) likewise emphasize the need for students to understand key AI concepts, competencies, and societal implications as part of computing education. Related research on machine learning instruction similarly points to the importance of helping students understand how models work, how they are developed, and how data-driven systems operate (Vartiainen et al., 2021). Based on these broader definitions of AI literacy, this approach is more closely aligned with contemporary AI literacy frameworks, which emphasize competencies such as understanding how AI works, evaluating and using AI, and attending to its ethical dimensions (Long & Magerko, 2020; Ng et al., 2021). It therefore raises an important question: What kinds of learning experiences best support students in developing an understanding of AI and its different dimensions?

To answer this question, one might first turn to K-12 standards. But AI literacy standards are only beginning to emerge and remain largely untested, and standards for data science education are themselves new. Still, the links between computer science, data science, and AI are unmistakable, and with data science education gaining momentum across the country, it offers a natural place for students to engage with the conceptual underpinnings of data and AI.

Understanding AI requires familiarity with how data are collected, labeled, curated, represented, and interpreted, because these processes shape how AI systems function. Preparing students to understand AI includes not only teaching AI-specific concepts, but also helping them engage with the practices through which data-driven systems are developed and evaluated. Data science education offers one way to make this possible.

2.2. Data Science as a Pathway to AI Literacy

Before continuing, it is important to define “data science.” Here, we draw on the Network of the National Library of Medicine’s definition of data science. Specifically, data science includes investigating questions with data by drawing on statistics, computer science, programming, and domain knowledge to collect, process, and analyze data in order to generate knowledge or solve problems (National Library of Medicine, 2026). If K-12 education approaches AI as a class of systems grounded in data, rather than only as a set of tools or concepts, data science education becomes an avenue for helping students develop deeper and more experiential understandings of AI. Whether in computer science classes, mathematics classes, or other instructional contexts, data science can engage students directly in the data practices that shape how contemporary AI systems work.

This connection is especially important as computer science education scholars argue that traditional programming-centered frameworks need to expand to better account for data and machine learning. Tedre et al. (2021), for example, argue that computational thinking should be reconsidered to place greater emphasis on machine learning and data, while Tedre and Vartiainen (2023) similarly argue that data literacy in the AI era should include what they call data agency, or the capacity for informed action, active control of information flows, and ethical data use. Wong et al. (2020) likewise point to the growing importance of data representation and intelligent systems in K-12 computing education. Lee (2025) positions data science as an emerging field at the intersection of statistics, computer science, and application domains. Together, this work suggests that data science can extend computer science education in ways that are more responsive to contemporary AI and the demands of AI literacy.

3. The Emerging Landscape of High School Data Science Education

In developing this argument, we draw on literature in AI literacy, computer science education, and data science education, as well as our prior research on high school data science curricula and implementation in Utah and Virginia.

3.1. Growth and Framing of High School Data Science

While statistics and data have long had a place in secondary education, data science has recently emerged as its own subject area, reflecting growing recognition of the role of data in society, work, and public life (Lee, 2025). This shift also reflects broader concerns about career readiness and civic participation in a world increasingly shaped by data and AI. These concerns also emerged in our research on high school data science implementation across two states and multiple districts. Teachers emphasized that students need to understand and work with data before entering the workforce, and that these skills can open pathways across a wide range of careers, whether or not students plan to pursue higher education (Schiffman & Century, 2024; Schiffman et al., 2026).

Labor-market trends reinforce this assertion. The World Economic Forum (2025) identifies big data specialists, AI and machine learning specialists, data warehousing specialists, and data analysts and scientists as among the fastest-growing jobs globally. The same report also identifies analytical thinking, technological literacy, and AI and big data as some of the core skills employers increasingly value. Salerno and Steemers (2024) likewise find that data-related skills are in demand across sectors, underscoring that preparation to work with data and AI is becoming broadly relevant, rather than limited to specialized occupations.

High school data science is beginning to take shape in multiple ways, including through standalone courses and approaches that integrate data science into existing courses (Lee, 2025). This emergence is significant because formal learning opportunities related to data have historically been limited—an estimated 75% of students will not have taken a probability or statistics course by the time they graduate from high school (National Center for Education Research, 2021). Sukol (2025) reports that 12 states and Washington, D.C., have added data science to their state course catalogs, 14 states have implemented data science teacher professional development, course pilots, or course sequences, and 7 states have adopted data science standards or frameworks. The same report estimates that in 2025, more than 70,000 students were enrolled in data science classes or in courses with integrated data science content—indicating that high school data science is already reaching students at a meaningful scale (Sukol, 2025). Together, these developments suggest that high school data science education is moving beyond local experiments and taking shape as a broader, though still uneven, national trend.

Despite the rapid development of data science in high school education, the aims and focus of instruction vary across states, schools, and curricula. These differences reflect broader variations in how data science is framed, with some approaches emphasizing programming skills, others focusing on statistical concepts, and still others prioritizing data cleaning or visualization. Lee (2025) describes this distinction as teaching data science “as process” versus “as content.” Teaching data science as process emphasizes investigation and the iterative practices of collecting, manipulating, analyzing, interpreting, and reporting data. By contrast, teaching data science as content emphasizes the statistical knowledge needed to understand data processes, the development of programming and coding skills, and, in some cases, more advanced topics such as machine learning and predictive analytics (Lee, 2025).

3.2. Data Science Implementation Across Curricula and State Contexts

Our prior research provides a concrete view of how these process- and content-oriented emphases take shape in curricula and state-level design. As part of this research, we analyzed four widely used high school data science curricula—Bootstrap, CourseKata, IDS, and youcubed—and found that both emphases are reflected in existing instructional materials, although they are emphasized differently across curricula (Schiffman & Century, 2024). All four curricula were aligned with the Common Core State Standards for mathematics, and two were also aligned with computer science standards—Bootstrap with the Computer Science Teachers Association Standards and youcubed with the California computer science standards. All four were designed to support data science instruction at the high school level as standalone courses, and some also provided modules, lessons, or units that could be integrated into existing courses. At the same time, the curricula differed significantly in their use of programming tools and technical platforms, as well as in their content emphases. For example, IDS and CourseKata rely on R, whereas Bootstrap uses Pyret and youcubed uses Python. Across the four curricula, the degree of emphasis placed on statistical modeling, programming and computation, ethics, social impact, and responsible data use varied considerably (Schiffman & Century, 2024).

Our research also examined how Utah and Virginia introduced data science in their high schools. In Utah, teachers selected from four common curricula that helped shape the course, whereas in Virginia the state first developed data science standards and teachers then built their own materials within that framework (Schiffman & Century, 2024; Schiffman et al., 2026). In both states, data science was positioned within existing mathematics pathways and counted toward graduation requirements, but the course was organized differently across the two state contexts.

Because data science draws on mathematics, computer science, and domain-specific inquiry, it can be taken up in different ways across schools and systems (Lee, 2025; Tedre & Vartiainen, 2023). Some implementations emphasize programming and computation more strongly, while others foreground statistical reasoning, real-world data investigation, or ethical and social questions related to data use. Together, these findings underscore that data science implementation is shaped not only by curriculum design or state policy, but also by the interdisciplinary nature of the field itself.

4. Implications for Access, Teacher Capacity, and Computer Science Education

4.1. Student Access

In the context of data-centered learning as a step toward AI literacy, one essential question to address is who gains access to that learning. Our research suggests that access depends not only on whether data science is offered, but also on how it is positioned within students’ broader programs of study. Course sequencing, prerequisites, graduation requirements, advising, and school-level messaging can all influence whether students encounter data science as an accessible and viable option. These structural factors matter because access to advanced and college-preparatory mathematics is already unevenly distributed, varying substantially by school characteristics and by students’ race, language status, and socioeconomic background (National Academies of Sciences, Engineering, and Medicine, 2024; Steiner et al., 2024).

Our research in both Utah and Virginia illustrates this dynamic. In both states, data science was introduced within mathematics pathways and positioned as an alternative to more traditional third-year high school mathematics, but the extent to which the course remained broadly accessible depended on local structures and decisions. In Virginia, for example, local decisions about prerequisites and course placement shaped whether data science was broadly accessible or available only to particular groups of students (Schiffman et al., 2026). In Utah, teachers similarly noted that students could not always take data science and remain on a pathway toward senior-year calculus, which affected how the course was perceived and who was encouraged to enroll (Schiffman & Century, 2024). In both cases, access depended not only on whether data science was offered, but on how it was positioned within broader school structures.

Whether data science functions as an alternative, a supplement, or a gateway to later coursework depends in part on decisions about placement, prerequisites, advising, and graduation credit. If schools want data science to broaden access to CS- and AI-related learning, they must attend to these institutional conditions alongside curriculum and instruction, given persistent disparities in who accesses and enrolls in computer science coursework (Weismann et al., 2024).

4.2. Teacher Capacity

Another consideration is teaching capacity. In both Utah and Virginia, high school data science was taught primarily by mathematics teachers, many of whom had limited prior experience with programming, let alone teaching programming. As a result, some teachers felt underprepared to support students during coding-related portions of the course, and some chose to limit or avoid those components altogether (Schiffman & Century, 2024; Schiffman et al., 2026). This pattern reflects a broader challenge in computer science education, where teachers are frequently recruited from other content areas and report knowledge gaps as well as a need for sustained content and pedagogical support (Zhu & Wang, 2024). Systematic reviews of computer science teacher professional development likewise conclude that, although such programs can improve teaching competence, they are often insufficient on their own to fully prepare teachers who enter computing instruction from other disciplines (Ma et al., 2025).

Both states provided meaningful teacher supports, but those supports were not sufficient. In Utah, teachers relied heavily on curriculum developer-provided professional development to learn the tools, technologies, and instructional materials associated with the course, while in Virginia, teachers benefited from standards-based guidance and statewide professional learning but still had to build and adapt substantial portions of instruction locally (Schiffman & Century, 2024; Schiffman et al., 2026). Even with these supports, many teachers felt they needed additional preparation to teach confidently across mathematics, data science, and computing.

If data science is to serve as a bridge to AI literacy, investment in teacher preparation must grow with it. Even skilled mathematics teachers need sustained support to teach across data science and its connections to AI, both of which continue to evolve quickly. That support works best when it reflects that teachers come to this work with different strengths—some in statistics, some in programming, some in project-based instruction, some in connecting with students. High-quality support meets teachers where they are, continues past a single workshop, and gives them time to plan alongside colleagues in mathematics and computer science. Teacher professional learning, however, has long been short on funding, time, and resources at both state and local levels, and building this capacity school by school is not realistic. With AI advancing quickly and students engaging with it daily, policymakers and administrators cannot kick teacher preparation down the road. Without that investment, the case for data science as a route to AI literacy remains an argument on paper rather than something students experience.

5. Extending Computer Science Education in an AI Era

The growth of high school data science has important implications for how students encounter computing in secondary education. Data science gives students opportunities to work with ideas and practices that overlap substantially with computer science. It does so through inquiry, interpretation, and real-world application across a range of subject areas. In data science, students may learn to work with data, build and interpret visualizations, use computational tools, and reason about how data and models shape conclusions. Such experiences can broaden the ways students encounter computing concepts by grounding them in data-centered investigations and applied questions.

This matters because many students already encounter AI on their own—through search, social media, and chatbot tools—usually without any framework for understanding how these systems work. Students need deliberate instruction in how data shape the systems they interact with, how models produce outputs, and why those outputs require interpretation and judgment. This need is heightened by a growing body of research suggesting that routine reliance on AI tools can encourage cognitive offloading and reduce the effort students invest in their own reasoning (Risko & Gilbert, 2016), with heavier AI dependence associated with lower critical thinking (Tian & Zhang, 2025). Although this work is still emerging and largely correlational, it also suggests that students who learn to question and verify AI outputs are better able to sustain their critical thinking, pointing to the value of learning experiences that build these habits.

Data science is especially well positioned to support this kind of learning because it places data at the center of inquiry. Rather than treating AI as an isolated topic to add to an existing course, data science gives students practice with the processes through which many contemporary AI systems operate. In tracing how data are collected and curated, interpreting model outputs rather than accepting them, and reasoning about the consequences of data-driven decisions, students build the kinds of evaluative habits this research identifies as protective. And because data science is inquiry-driven and open-ended, students can use it to investigate and interpret, moving beyond consuming AI to asking their own questions of it.

Data science education is not the only way to develop AI literacy. Some suggest introducing AI as explicit subject matter (Touretzky et al., 2019), while others embed AI topics within computer science courses. Others focus on responsible use, privacy, and the evaluation of AI tools students already encounter. While these approaches are important and complementary to data science, data science offers students the opportunity to engage directly in the data practices and processes that underlie AI. This is why, in the age of AI, foundational understanding of AI may be built not only in standalone computer science courses but also in data science and other interdisciplinary, data-centered settings. If students are to make sense of AI systems in informed and critical ways, then computer science education should include greater recognition of learning that happens through data-centered work. In this sense, data science should be understood as an important part of the broader educational landscape through which students develop the knowledge and habits of mind that understanding AI now requires.

6. Conclusions

AI is reshaping what computer science education needs to accomplish. As AI systems become more embedded in everyday life, work, and schooling, preparing students to understand them requires more than a traditional emphasis on algorithms and programming alone. It also requires attention to data and to the practices through which data are collected, interpreted, and used in the systems that increasingly structure social, educational, and professional life. In this context, data science offers an important way for students to develop a foundational understanding of AI by connecting computation, interpretation, and context through inquiry with data.

At the same time, the emerging landscape of high school data science makes clear that this potential will not be realized through curriculum alone. Whether data science can serve as a meaningful route into AI literacy depends on the structures that shape student access, the preparation and support available to teachers, and the extent to which schools recognize interdisciplinary forms of learning as critical to computing education. In an AI era, the importance of data science lies in expanding the range of educational settings through which students can encounter and make sense of computational ideas. Recognizing data science as an important part of that broader landscape may help schools respond more thoughtfully and effectively to the demands AI now places on education.