Slanted Speculations:

Envisioning Narrative Intelligence: A Creative Visual Storytelling Anthology

“…To address narrative biases in visual storytelling, we do not call for "solving" them. Instead, we look to responsibly generate creativity from them, in line with scholars such as Benabdallah et al whose work on bias makes the case for "not resolving it but rather tilting it so that it can cast a different light, generate different encounters" [11]. Further, along with Cambo and Gergle, we call for ongoing recognition of biases by adapting concepts of positionality and reflexivity to data science and ML [19] for computational visual storytelling in particular.…”

“…More broadly, HCI scholars have shown how stories are told and whose stories are told in the visual language of textiles, for instance, can either work to elevate marginalized histories, or marginalize histories such as racial and socioeconomic origins [26], as well as gendered legacies [70]. Hence, to reorient the false sense of "objectivity" in dominant HCI methods, Rosner posits the concept of "critical fabulations" as narrative tactics for reworking design [69] and its histories, while also working with Benabdallah et al to recognize the generative potential of harnessing biases in creative, yet critical ways [11]. With that said, irresponsibly imbuing human biases in algorithms can harmfully reproduce and scale them [13,38], particularly along the axes of race and gender surfaced in computerbased vision tasks [18].…”

“…On one hand, this has potential to generate narratives that advance harmful prejudices about people. But on the other hand, inevitable biases may be generative when explored critically and creatively [11]. Given this paradox, before computerizing stories with potentially hazardous systems (as prior work on algorithmic oppression [61] and biases in computer-based vision tasks [18] have shown), we aim to first understand the nuances and risks that arise when people (never mind machines) undergo a systematic process of deriving stories from image sequences.…”

Folklore at the University of Oregon:

“…Drawing from speculative approaches, this work has sought to use artistic and exploratory experimentation with algorithmic systems and related data practices to probe the limits and potentialities of "bias, " material and cultural phenomenon. In a series of works [8,49,69,70], Gabrielle Benabdallah, Chari Glogovac-Smith, Afroditi Psarra, Caitlin Lustig and several others have examined "bias" through textile etymologies, where the term refers to the angular cross grain of woven fabric. With exercises such as "cutting on the bias" [69] (the diagonal) of fabric to create a data set, they examine how the production of data requires making choices about where to make cuts or lines between what is in and out-reflecting both the environment and the dataset creator (their assumptions, skills, tools etc).…”

“…5.1.1 Design Speculation 1: Lacey Jacoby's AI Translation. Our first example of translation work comes from design researcher Lacey Jacoby [8], who reimagined Google Translate to reveal linguistic cultural bias often hidden within the tool. "Welcome to a new translation experience" the app announces.…”