AI for Not Bad

Abstract

Hype surrounds the promotions, aspirations, and notions of "artificial intelligence (AI) for social good" and its related permutations. These terms, as used in data science and particularly in public discourse, are vague. Far from being irrelevant to data scientists or practitioners of AI, the terms create the public notion of the systems built. Through a critical reflection, I explore how notions of AI for social good are vague, offer insufficient criteria for judgement, and elide the externalities and structural interdependence of AI systems. Instead, the field known as "AI for social good" is best understood and referred to as "AI for not bad."

How to Design AI for Social Good: Seven Essential Factors

“…Respect for privacy is also a necessary condition of human dignity, since we can view personal information as constituting an individual, and deprivatising records without consent is likely to constitute a violation of human dignity (Floridi 2016). The conception of individual privacy as a fundamental right underlies recent legislative action in, for example, Europe (through its General Data Protection Regulation) and Japan (through its Act on Protection of Personal Information), as well as judicial decisions in jurisdictions such as India (Mohanty and Bhatia 2017). Privacy supports people in deviating from social norms without causing offense, and communities in maintaining their social structures, so privacy also undergirds social cohesion.…”

“…Approaching AI4SG ad hoc, by analysing specific areas of application-like famine-relief or disaster management-as an annual summit for AI industry and government has done (ITU 2017(ITU , 2018; "AI for Good Global Summit" 2019) indicates the presence of a phenomenon, but neither explains it, nor does it suggest how other AI4SG solutions could and should be designed to harness AI's full potential. Furthermore, many projects that generate socially good outcomes using AI are not (self-)described as such (Moore 2019).…”

Editorial: Shaping Ethical Futures in Brain-Based and Artificial Intelligence Research

“…When and how is AI autonomous, and what are the characteristics of autonomous AI? The discussions around this topic range from more technical notions as to what counts as "autonomous" in terms of the need for human intervention (such as the different levels of vehicle autonomy in autonomous cars) to wider questions of machines as subjects of autonomy (SAE 2018;Moor 2006). What will an autonomous AI (in the philosophical rather than the technical sense) look like and what does this mean for its interactions with us?…”

‘AI for Social Good’: Whose Good and Who’s Good? Introduction to the Special Issue on Artificial Intelligence for Social Good

“…The twin aims of this Special Issue, simply stated, are to interrogate the plausibility of this notion and to consider its implications. The case that AI may -if developed carefully and deployed sensitively (Floridi et al, 2020) -contribute to net-positive outcomes in (some) socially relevant spheres is not without foundation. An array of efforts and initiatives are already underway, for example, to develop AIbased responses to help meet the UN's Sustainable Development Goals (SDGs) by 2030 (Cowls et al, 2021;Vinuesa et al, 2020).…”

“…Such questions are not novel. Moore (2019), for example, has explored the meaning and implications of 'AI for social good' and argues instead for 'AI for not bad', citing the vagueness and critical inadequacy of the term, bringing to mind Taylor's (2016) question regarding whose good is meant to be served by the use of big data as a 'public good'. But addressing these questions more fully -about whose good is served by, and who ought to be thought of as good as a result of, AI for social good -benefits from the series of empirically and ethically grounded contributions that are assembled in this Special Issue.…”