Paper Short Abstract:

Despite their abundance, images found in social media or historical archives often lead to long paths to access or consent, hybrid heterogeneous datasets, laborious organizing and coding tasks, and ethical dilemmas to publishing. Akin to the topic of “unwriting”, this paper examines these research practices to invigorate the role of qualitative scholars’ “looking practices” in the age of AI

Paper Abstract:

This presentation is based on an ongoing study about the challenges social, historical, and cultural researchers face when working with the abundant but fragmented imagery of past and present. The material comes from a multi-researcher project (Late et.al. 2024, forthcoming) involving 21 in-depth interviews conducted in 2024, capturing scholars’ experiences with visual sources. Participants, mostly qualitative researchers, highlighted challenges such as complex pathways to finding data, laborious collection of diverse materials, personal organizing methods, and difficulties in publishing. Their often undocumented “practices of looking”—described by Sturken and Cartwright (2009) as active and purposeful acts of decoding and coding of meanings— reveal some obvious challenges but also less explored opportunities to participate in a fast-paced technological landscape.

There are known ethical and technological challenges that hinder researchers from fully embracing digital tools; such as GDPR, restricted access to social media platforms, limited contextual information about material, topped by a general feeling that AI technologies are still a “long way” to support their qualitative approach to research. Despite these obstacles, researchers have grown accustomed to abundance through refined apprising criteria, cumulative collecting and engaging in digitization, although systematic annotation and preservation strategies remain lacking. This reinforces a need for a broader, “bird’s-eye” perspective, which is more common in 'hard' digital humanities approaches.

As a way forward, I propose recognizing researchers' meaningful groups of images as valuable assets that can help develop infrastructures and AI applications that "learn" to alleviate the task of making sense of today’s visual and ever complex research landscape.