At the Tensions of South and North: Critical Roles of Global South Stakeholders in AI Governance

A Human Rights-Based Approach to Responsible AI

“…Both approaches are problematic given the wide variation in the moral beliefs people actually hold. 6 Indeed, since most FATE research in AI is situated in the West, it does not meaningfully engage with the conditions, values, and histories of non-West contexts [41,57,66]. As Birhane and Cummins note in [8], "it is possible that what is considered ethical currently and within certain domains for certain societies will not be received similarly at a different time, in another domain, or for a different society. "…”

Identifying Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction

“…Cultural harm has been described as the development or use of algorithmic systems that affects cultural stability and safety, such as "loss of communication means, loss of cultural property, and harm to social values" Macro socio-economic harms "Harms associated with the labour and material supply chains of AI technologies, beta testing, and commercial exploitation. " [151] Environmental harms "The energy cost of training machine learning models... [and] harms from intensive water and fuel usage and server farms, consequent chemical and e-waste. " [151] (p. 30) [4].…”

“…[151] Environmental harms "The energy cost of training machine learning models... [and] harms from intensive water and fuel usage and server farms, consequent chemical and e-waste. " [151] (p. 30) [4]. As algorithmic technologies can "foreclose alternative ways of understanding the world and restricting imaginations about possible futures" (p. 162) [164], the nature of their harm can encompass adverse cultural such as systemic erasure [61], eurocentric ideas being exported to Global South [67,129], harmful cultural beliefs [66], such as normalizing a culture of non-consensual sexual activity [123], or proliferating false ideas about cultural groups [68,167].…”

“…Algorithmic systems can increase "power imbalances in socioeconomic relations" at the societal level [4,118] (p. 182), including through exacerbating digital divides and entrenching systemic inequalities [96,204]. The development of algorithmic systems may tap into and foster forms of labor exploitation [67,129], such as unethical data collection, worsening worker conditions [25], or to technological unemployment [45], such as deskilling or devaluing human labor [151]. For instance, text to image models may undermine creative economies [208].…”

“…Environmental harms entail ecological harms, such as the depletion or contamination of natural resources [23,81,120,129,181,182,208], and damage to built environments [120]. Ecological harms concern adverse changes to the "ready availability and viability of environmental resources" (p. 738) [126] that may occur throughout the lifecycle of digital technologies [151,210] from "cradle (mining) to usage (consumption) to grave (waste)" (p. 169) [23]. Similar to other sociotechnical harms, the "benefits and burdens of extractivism are unevenly distributed around the planet" whereby consumption in the economic core are contingent on extraction from the economic periphery [23] (p. 170).…”