Can Artificial Intelligence Be an Ethical Collaborator in Educational Research?

• In Event: Dreams for Digital Spaces: What Shapes the Worlds of Children, Educators, and Researchers?

Abstract

We invite the audience to consider the affects and effects of algorithms and algorithmic infrastructures on digital educational research by via data visualisations generated while play-testing AI-based social listening tool, SentiOneTM Positioning SentiOneTM as collaborator, our interdisciplinary team explores the feasibility of a hybrid computational-human approach to performing nuanced sentiment analysis with large digital datasets. Conducted with a view to tracing the changing dispositions (Threadgold, 2020) towards children’s use of digital technologies in the home since March 2020, our explorations were underpinned by a need to know the extent to which social listening may be incorporated into research seeking to map the affects and emotions that run through dynamic, digital archives such as Twitter.

The presentation takes in two key concepts, 1) becoming-with AI and 2) data in/justice; with both being central to the feminist materialist thinking which frames the larger project and generating data through digital methods (here social listening). We argue that acknowledging the productive force of SentiOneTM’s algorithms and intentionally ‘becoming-with AI’ fosters ethically affirmative data engagements characterised by criticality and creativity. Conceptualising AI-based research thus contributes to data feminisms scholarship (D’Ignazio & Klein, 2020) by providing a way to avoid algorithmic imperialism involving extractive, fixed, normative data practices that position algorithms as “digital workhorses of the economy” (Goodman, 2020, p.50).

Given the potential of AI-based social listening tools like SentiOneTM for transforming how data is gathered and analysed for educational research purposes, consideration of algorithms and algorithmic infrastructures is crucial to ensuring data justice occurs when AI is invited into a research collaboration. Further, we propose a hybrid computational-human method forforfor working with AI that offers a computational attunement to affect and emotion.

Authors

• Sarah Healy, University of Melbourne
  Presenting Author

• Amanda Belton, University of Melbourne
  Presenting Author

• Karen Thomson, Melbourne Data Analytics Platform, The University of Melbourne
  Non-Presenting Author