ECPR

Install the app

Install this application on your home screen for quick and easy access when you’re on the go.

Just tap Share then “Add to Home Screen”

ECPR

Install the app

Install this application on your home screen for quick and easy access when you’re on the go.

Just tap Share then “Add to Home Screen”

Democratising AI? Big AI and Barriers to the Democratisation of AI Production Networks

Political Theory
Normative Theory
Technology
James Muldoon
University of Essex
James Muldoon
University of Essex

Abstract

In recent years there have been a number of proposals for the democratisation of the digital economy through various forms of social ownership over digital infrastructure. This paper analyses the potential of these proposals in the era of Big AI, noting three new significant barriers to this vision, related to a further concentration of corporate power, geopolitical rivalry, and supply chain dynamics. First, the evolution of Big Tech into Big AI underscores the intensifying concentration of power in a handful of new tech empires. Building and scaling AI systems require immense capital investments, far exceeding those seen during the platform era. This capital intensiveness restricts the development of AI to legacy tech giants and their aligned startups, consolidating control over critical infrastructure, talent, and datasets. The monopolistic tendencies of Big AI create an environment where smaller firms, worker cooperatives, or public interest initiatives face near-insurmountable barriers to entry. Consequently, the promise of democratisation remains elusive, overshadowed by the entrenchment of a few entities in positions of disproportionate influence over tech development and public policy. Second, the geopolitical rivalry between the United States and China further constrains the potential for AI democratisation. National governments, motivated by global competition, are hesitant to regulate or ‘break up’ AI firms, fearing economic and strategic disadvantage relative to their superpower adversaries. This competition perpetuates a narrative of AI development as a national security imperative, where the dominance of domestic tech firms is seen as a geopolitical asset. As a result, governments prioritise industrial and technological advancement over social or ethical considerations, sidelining efforts to decentralise power within the sector. Third, the globalised and fragmented nature of AI supply chains exacerbates the challenges of improving working conditions or fostering alternative forms of production. Data annotation, content moderation, and other forms of human labour underpinning AI are often outsourced to regions with minimal labour protections. Competitive pressures drive a race to the bottom in wages and working conditions, leaving worker-owned cooperatives or ethical firms unable to compete with low-cost providers. The razor-thin margins of this labour market further entrench exploitative practices, as ethical alternatives are priced out of the global marketplace. Advocates for democratising the digital economy must face these new realities and envision forms of democratisation that takes these concerns seriously. This analysis contributes to critical discussions on AI governance, emphasising the urgency of systemic reform to counteract the inequities entrenched in contemporary AI production networks.