AI Education for the AI Generation: A Study of Computer Science Student Attitudes Towards AI Ethics and Implications for CS Curricula

Abstract

The proliferation of artificial intelligence (AI) technologies in recent years is evident across a variety of domains and settings, from AI voice assistants like Siri to an AI overview on Google searches to Waymo self-driving cars. Computer science (CS) students studying AI therefore find themselves at the forefront of learning about and using cutting-edge AI technologies. Their dual roles as AI learners and AI consumers place them in a unique position to influence the development of policies to govern the ethical use of AI. Yet their use of these tools and attitudes towards AI are under-studied. In order to strengthen the CS curriculum to prepare students for an AI-centric future, we must first explore how students think about and interact with AI.

To address this gap in the literature, this thesis studies CS student attitudes and competencies with respect to AI and AI ethics, seeking to understand how students' learning about and use of AI influences their opinions on issues of social impacts and regulation of AI. We situate our work at the crossroads of research on AI ethics, AI policy, and CS education, and we employ a mixed methods approach to investigate CS student use of AI tools, attitudes towards AI ethics, and knowledge of the AI policy landscape. Our findings were used to motivate the development of a two-lecture "AI Policy Module" aimed at familiarizing students with ethical impacts of AI and improving student awareness of and engagement with the emerging landscape of AI policy. This module was successfully piloted in an introductory graduate machine learning course.

This research advances understanding of this population of 'AI pioneers', identifying three themes that frame CS students' unique engagement with AI: widespread use of AI as a "tool"; awareness of but indifference towards the ethical harms of AI systems; and distrust of and disengagement with the policymaking institutions. Throughout this thesis, we build on our novel conceptualization of the dual roles of 'AI students' as learners and consumers, and emphasize the pedagogical as well as political implications of our findings. We conclude by summarizing the important contributions of this research, and by identifying directions for future research, including the development of a full-semester course on AI policy and the in-depth exploration of political engagement among this young generation of AI users.