CAPE for AI: A New Framework for Navigating AI in Education

      When computing education pioneer Jane Margolis and colleagues wrote Stuck in the Shallow End in 2008, they used the history of segregated public swimming pools to illustrate how systemic barriers kept Black and Brown students out of core computer science education. Today, the rapid proliferation of generative AI has presented a very different challenge: students and educators have been abruptly thrown into the deep end without swim lessons, lifeguards, or clear rules for the water. 

      While early digital divides centered on who had access to computers and the internet, the current AI divide is far more complex. It impacts student agency, critical thinking, and socio-emotional development. Moving from basic access to true AI literacy requires moving past high-level guidance focused on academic integrity or tool usage to evaluating how technology actually affects student development. 

      To help education systems navigate this shift, we introduced CAPE for AI at the recent ACM AI Leadership Summit - a practical evolution of the widely adopted CAPE Framework (Fletcher & Warner, 2021). Like the CAPE Framework, CAPE For AI provides K-12 leaders, educators, and researchers with a structure for formulating guiding questions for AI education (AIEd) in four dimensions: Capacity, Access, Participation, and Experience. Through these questions, K-12 ecosystems can develop and evaluate holistic AI strategies that support AI literacy for all educators and students, truly preparing them for the challenges of an AI-infused world.

      CAPE for AI is distinguished from the original CAPE Framework in several key ways. Whereas traditional CS education uses clear data markers like course enrollment numbers, dedicated course codes, and specific teacher training certification, AI education will likely operate differently in part due to the ubiquitous access that many students already have to AI tools outside of school. Some key considerations need to be taken into account to develop practical strategies for addressing AI education and for researching the impact of AI on K-12 students. 

      • Cross-Disciplinary Integration: AI may be less likely to be taught as a standalone subject, as compared to CS or core programming courses, particularly at the secondary level.  Because AI usage is already evident across disciplines, literacy should be embedded across English, math, history, and the arts.
      • Beyond Course Access: Equity in AIEd can no longer be measured by who has access to a tool, but by how ubiquitous access impacts students and educators. 
      • Human and Psychological Impact: AI usage by students without guardrails risks cognitive offloading, diminished student self-efficacy, and widened equity gaps if human relationships are swapped for automated tools. This threat to students could be exacerbated in under-resourced schools who lack the bandwidth to develop and implement a comprehensive strategy for addressing or evaluating student AI usage on their own.

      In short, the ethical implications of AI on individuals and society have quickly become a primary consideration for educators (both in K12 and higher ed), students, policymakers, parents, and society at large to a greater degree than what has been previously seen in computing education.  CAPE for AI can be a tool for interrogating those ethical implications and human impacts, providing a lifeline for educators, researchers, and policymakers to keep their head above water as our society grapples with the rapid pace of AI integration.

      CAPE for AI framework showing four dimensions built on human impacts: Capacity, Access, Participation, Experience.
      This is a sample of questions that should be asked when interrogating AI education infrastructure.

      Text Description of the CAPE for AI Infographic

      CAPE for AI examines AI education infrastructure through four connected dimensions built on a foundation of human impacts.

      Capacity: Teachers, Funding, and Policies

      • How are educators prepared and supported to address AI’s social, cognitive, and psychological impacts on students?
      • Do policies and learning standards address the technical and ethical implications of AI in education?
      • How should educator AI literacy be assessed?

      Access to Instruction

      • How must instruction change in every subject to account for ubiquitous student AI access?
      • Do educators have clear criteria for evaluating and selecting AI tools for teacher and student use?

      Participation in Instruction

      • How do we measure authentic learning in AI-assisted instruction versus cognitive offloading?
      • Are there disparities in which students are assigned to AI-led versus teacher-led instruction?

      Experience: Student Outcomes

      • How does AI-assisted instruction affect students’ psychological development?
      • What evidence shows that AI-assisted instruction improves student learning?

      Fletcher, C. L., & Warner, J. R. (2021). CAPE: A framework for assessing equity throughout the computer science education ecosystem. Communications of the ACM, 64(2), 23-25. 

      Margolis, J., Estrella, R., Goode, J., Holme, J. J., & Nao, K. (2008). Stuck in the shallow end: Education, race, and computing. The MIT Press.

      McCune, J., Zarch, R., Fletcher, C., & Childs, J. (in press). From stuck in the shallow end to thrown in the deep end: What every student needs to thrive in an age of AI. Proceedings of the ACM AI Leadership Summit.