In recent years, AI for Good (AI4G) – or also AI for Social Good (AI4SG) – has become a catchy label for initiatives, gatherings and funding calls, all revolving around the potential of Artificial Intelligence (AI) and Machine Learning (ML) to achieve the Sustainable Development Goals (SDGs).
This conversation, shaped by win-win narratives on the synergies between business and development/humanitarian stakeholders, has been mainly animated by technological corporations, global consultancies, and international organisations, and has coalesced around a plethora of publications and events bringing together corporate executives, start-uppers, policymakers and development practitioners.
Big Tech like Google and Microsoft, just to name the most prominent ones, have translated this AI-driven commitment to the SDG agenda, for instance, into platforms to prevent and mitigate flooding, or databases to organize and make available petabytes of environmental monitoring data.
AI is not always for good
However, these initiatives have so far yielded mixed results. Reviews of AI4SG projects show their uneven distribution across the SDG agenda, with the overwhelming majority of projects addressing SDG 3 (‘Good Health and Well-Being’).
Other studies have suggested that the over-reliance on AI systems could contribute to the reproduction of structural inequalities and injustice built into the data sets used to train predictive and generative models. This risk is likely to be compounded by the cost-cutting logic underpinning the data collection and annotation processes that are central to ML and by the opacity of the systems, which hinders accountability.
AI is driven by Big Tech
Moreover, the popularity of AI since early 2010’, when the boom of the neural networks, or deep learning, approach has rekindled the interest of the industry, has coincided with the increasing influence of the tech industry over AI research and ethics agenda through massive hiring of AI scientists, computing power, and large datasets.
Big Tech in particular has positioned itself as the main force shaping the research trajectory and the policy and popular conversation around AI by leveraging not only technical and financial resources, but also its geopolitical clout.
While the academic attention on the political economy and policy relevance of Big Tech is growing, the implications of Big Tech’s influence on the global development arena have hitherto been largely overlooked. Particularly relevant for this blog is the currently limited discussion on the continuity, discontinuity and tensions between ICT4D/DD and AI4SG.
Does AI4G equal ICT4D?
Is ICT4D surreptitiously being subsumed by AI4SG, or is it struggling to keep its own identity?
Since ICT4D 1.0, and in the wake of the post-2015 agenda, which placed greater emphasis on the private sector as an agent of development, tech companies have become more prominent in the development space.
With the rise of AI, corporate actors have taken the lead in going beyond siloed solutions to developmental challenges and focused instead on the expansion of data infrastructures in the Global South – the very infrastructures producing evidence on which policies are based. They have built their hegemony not only upon their technological supremacy, but also on crafting an AI-driven development discourse that has been largely legitimized by policymakers, donors, development practitioners, and, to some degree, academics.
And yet, as suggested by critical humanitarianism scholars, AI4SG is grounded in a culture of “humanitarian neophilia”, in which an “optimistic faith in the possibilities of technology” is combined “with a commitment to the power of markets”. Through a “top-down approach that presupposes what good is”, AI4SG attempts to deflect criticism by spilling over into a moral sphere and transcending the domain of development and, eventually, politics.
Focus on development
ICT4D can differentiate itself from AI4SG by continuing to put stress on the D of Development and keeping the spotlight on the political nature of the concept – a nature that the SG of Social Good is concealing behind a technocratic veneer.
This is a conversation that ICT4D scholars and practitioners need to have sooner rather than later.
A lightly edited version of AI for (whose) good? Implications for the ICT4D community by Gianluca Iazzolino
Extremely well said. The same argument has been made about the concept of ICT/Tech/Design ‘for good’ in general, with the big difference that, as you point out, big tech has a chokehold over the development of AI/LLMs, and any work our industry does to making these models more ethical and reliable for use in the global south will ultimately be benefiting shareholder accounts. But in the same way that we confronted our panic about [insert new tech here] I do see promising signs that we are making attempts to address these concerns, not just blindly implementing, for the most part. Something about the opacity of LLMs is making it necessary for implementers to pause and think in a way that didn’t happen as much with apps, for example.