Developing Artificial Intelligence Literacy by Examining the Politics of Internet Platforms

• In Event: Artificial Intelligence in Education: Potentials and Challenges in Ethics, Data Privacy, and Pedagogy

Abstract

Thanks to the extraordinary efforts of scholars, journalists, artists, and activist organizations, (e.g., Benjamin, 2019; Crawford, 2021; Eubanks 2019; Gillespie, 2019; Wells et al., 2022; Kantayya, 2020; Noble, 2018; Zubuff, 2018; the Algorithmic Justice League; Allied Media), there is a growing recognition that AI-powered systems (AIs), for all their benefits, can and do cause significant individual and societal harms – particularly for society’s most vulnerable – and therefore need better oversight and accountability. However, designing and enacting better models of AI governance requires enhancing public understanding of AI technologies, including their strengths, weaknesses, and vulnerabilities; and their capacity to shape our individual, social, and political lives.

In this symposium, we will discuss some approaches that we have taken in the XXXX lab at YYYY University to help high school and undergraduate students learn about AIs in the context of internet platforms. Internet platforms like YouTube, Instagram, and TikTok are not only ubiquitous and familiar, but they also rely heavily on AI to moderate and shape the global information and communication landscape. As such, they offer a rich, situated context for examining the technical as well as ethical and sociopolitical implications of AI. For instance, from a technical perspective, platforms could be analyzed in terms of how they collect data about their users, or how these data are used to classify, rank, moderate, and recommend content. Platforms could also be examined using a socio-political lens by looking at the role of incentives (e.g., user satisfaction, profit maximization) in shaping how content is prioritized and circulated; or how the racial and gender identities of users mediate platform experiences in consequential ways.

Drawing on data from two ongoing initiatives – an undergraduate tech ethics course, and an after-school program for high school students – we will describe and analyze three different approaches for helping novices learn about internet platforms:

The empirical approach – where learners directly observe/interact with the data and categories that shape their content/ad experience. For instance, by examining and discussing their own data profile (e.g., search and browse history, location data, and marketing and interest categories derived from this data) learners can gain new insight into how they are seen (or not seen) through the eyes of a commercial internet platform.

The cultural narrative approach – where learners read/watch and discuss stories of people who impact and are impacted by platforms (e.g., SciFi, critical news stories, academic articles, documentaries, leaks, etc.), and connect these stories to their own lives and experiences. This approach can help learners connect specific engineering and design decisions (analyzed via the “empirical approach”) to specific, material consequences in the real world.

The artifact approach – where learners synthesize their socio-technical understandings of platform AIs by creating public-facing artifacts of their learning (e.g., documentaries, podcasts, and educational modules) to help others’ better understand AIs, and some of the steps they might take to help build more equitable and just AIs.

Authors

• Sarah Van Wart, University of North Carolina - Asheville
  Non-Presenting Author

• Sepehr Vakil, Northwestern University
  Presenting Author

• Natalie Araujo Melo, Northwestern University
  Non-Presenting Author