Abstract:

International frameworks, such as a draft framework by the European Commission, OECD, and code.org, emphasize ethics as a constitutive dimension of AI literacy. Learners need to recognize that AI operates within social and political contexts and can reproduce inequities if its inputs and outputs, and human decision making at every stage of development and deployment, go unexamined. It not only involves questioning how training data are gathered and whose perspectives are represented or excluded, but also a consideration of sources of bias and how to mitigate them throughout the design cycle of AI systems. In this paper, we present a framework that an interdisciplinary AI Institute has developed to guide instructional designers and support educator learning for integrating ethics into problem-based units of instruction.

The units where we are applying the framework target AI literacy for secondary students, and are accompanied with educator professional learning materials. The framework has four core commitments that students learn about:
(1) participatory design;
(2) human-centeredness;
(3) values driven decision-making; and
(4) anticipation and evaluation of impacts.

In addition, the framework specifies ethical considerations at multiple phases of what we call the “Machine Learning Design Cycle,” spanning project and task conception, data collection, model development, documentation, feedback, and evaluation. A key contribution of this framework is its attention to documentation as a core ethical and pedagogical practice. Documentation, through dataset datasheets, model cards, design journals, ethical risk registers, and student reflections, supports transparency, accountability, and reproducibility while helping learners articulate the human decisions that shape AI systems. Treating documentation not as an afterthought but as a design value helps students recognize the epistemic labor involved in building responsible AI systems.

In our presentation, we will describe how the framework has been used to inform the design of a new unit focused on Generative AI and art, which focuses on teaching students about the technical aspects of AI image generation alongside the accompanying ethical dilemmas of intellectual property and creativity, and the redesign of a unit focused on human-AI teaming in moderating online communities. We present examples of specific activities that engage students in analyzing the intertwined technical and ethical issues that informed design decisions from real cases of ML system design, as well as activities that engage students in taking up these issues in the design of their own AI prototypes. In sum, the framework illustrates how integrating ethical reflection and documentation into the AI design process can help students understand the social and technical dimensions of AI systems and help to prepare them to engage responsibly in the governance of AI technologies.

Keywords:

AI literacy, ethics, participatory design.