It was in January 1945, as the Allies prepared for their final assault on Berlin and the shape of the post-war world hung in the balance at Yalta, that Vannevar Bush first articulated his vision of the "memex", a device that would augment human memory and transform how we process information. Bush could scarcely have imagined that eight decades later, his conceptual descendant would arrive not just as a tool but as an economic force; nor that its benefits would prove as unequally distributed as the colonial spoils his generation had inherited. The artificial intelligence market, projected to reach $4.8 trillion by 2033, promises productivity gains and digital transformation on a scale unprecedented in human history. It also threatens to entrench the very inequalities that successive waves of globalisation have failed to remedy.
And also, perhaps more troublingly, to create new ones.
The concentration of AI resources in wealthy nations and a handful of corporations represents technological enclosure. The digital equivalent of those parliamentary acts that fenced off common lands and created the preconditions for industrial capitalism. Consider the arithmetic. Forty percent of global corporate research and development spending in artificial intelligence is controlled by just one hundred firms, predominantly headquartered in the United States and China. The handful of companies producing critical AI hardware collectively command more than eighty percent of the global chip market, whilst the mining of essential raw materials like gallium, germanium, and their arcane cousins remains concentrated in a few economies.
Leading technology giants now possess market valuations that rival the gross domestic product of the entire African continent. A comparison that would have seemed fantastical a decade ago, obscene today, and possibly routine tomorrow.
This isn't simply about market capitalisation or the familiar dynamics of late capitalism. The monopolistic control over both physical infrastructure and intellectual development creates what the United Nations Trade and Development agency has termed a "winner-take-all dynamic," though the phrase rather understates the violence of the process. Each advantage in AI development generates network effects that make subsequent catch-up difficult and perhaps structurally impossible.
The warnings from international organisations carry the weight of precedent. The UN notes that AI-driven automation tends to favour capital over labour (as automation invariably has done since the Luddites first smashed their frames in Nottinghamshire), potentially widening inequality and eroding the competitive advantage that low-cost labour traditionally provided to developing economies. The implications extend beyond simple job displacement, though that prospect alone should give pause, with projections suggesting the technology could affect forty percent of employment worldwide. What's at stake is the fundamental architecture of economic development itself.
For nations that built their strategies around manufacturing and labour-intensive industries, the evaporation of comparative advantage represents more than cyclical disruption. It constitutes an existential challenge to their development models. Surveys indicate that forty-one percent of employers were planning workforce reductions in areas where artificial intelligence could replicate human tasks.
When manufacturing employment disappears in countries lacking robust education systems or alternative opportunities, the result is potential for social dissolution on a scale that should haunt policymakers in Geneva and New York, if such institutions still harbour the capacity for prescient dread.
But as these warnings remind us, technology's impact on inequality operates through multiple channels, not all of them obvious. There exists what might be called the governance gap: one hundred and eighteen countries, mostly in the Global South, remain absent from major AI governance discussions. This exclusion from policy-making tables means that the rules shaping AI's development and deployment are being written without input from the majority of the world's population.
This matters, and not just as a question of procedural fairness. AI systems trained primarily on data from wealthy nations demonstrate predictable biases when deployed in other contexts; facial recognition systems, for instance, have shown significant error rates on populations underrepresented in their training data. Without diverse voices in governance frameworks, AI development risks encoding the perspectives and priorities of a small segment of humanity whilst imposing those systems globally, or what researchers have begun to call, without irony or hyperbole, "technological colonialism."
Developing nations become consumers of technologies designed elsewhere, with little capacity to shape them to local needs or values, a dynamic that previous generations might have recognised from the history of railway concessions and mineral extraction rights.
The infrastructure divide tells a parallel story. The trade in physical AI-enabling goods like semiconductors, computers, and chemicals was worth $2.3 trillion in 2023, a figure that hints at but cannot fully capture the broader chasm separating wealthy and poor nations. Economic modelling by the World Trade Organisation suggests that if lower-income countries close the digital infrastructure gap by half (a qualification that itself speaks volumes), their income growth could reach fifteen percent by 2040 through wider AI adoption. Without such progress, these economies face income increases of only eight percent, compared to fourteen percent for high-income nations. The mathematics of divergence, expressed with the cool precision of econometric models.
This infrastructure divide encompasses more than fibre optic cables or data centres, though these too demand investment beyond the reach of many nations. It extends to reliable electricity, skilled workforce development, regulatory frameworks, and access to computational resources necessary for AI development and deployment. The costs are prohibitive. The expertise required extends far beyond simple technology transfer. Moreover, regulatory fragmentation creates additional barriers, with poorer countries often adopting the most restrictive policies from an abundance of caution or a simple lack of capacity to develop nuanced frameworks. This regulatory patchwork makes it harder for companies in developing nations to participate in global AI trade and innovation.
However, and happily, the tools to address AI inequality already exist. What's missing is political will rather than technical capacity. International organisations recommend several interventions that possess the virtue of pragmatic feasibility.
If we can translate the enormous economic value AI generates into shared prosperity through strategic investments, inclusive governance, and international cooperation, the technology could help reduce global inequality rather than exacerbate it. This future is possible, though far from assured. The window for action narrows as AI systems and governance structures solidify. The choice between AI as a force for global equity or another mechanism for concentrating wealth and power remains ours to make, though perhaps not for much longer. That redistribution must also reach beyond industrial productivity into the social and cultural domains of AI. The availability of AI companions, creative tools, and personal assistants illustrates how access to artificial intimacy, knowledge, and creativity has itself become a marker of privilege. Equalising participation in these softer, human-facing branches of AI is as urgent as balancing the harder economics of chips and data centres.
One thinks of Vannevar Bush again, imagining his memex in that vanished world of 1945, and wonders what he would make of what we've built, and what we've failed to build. The stakes for billions of people demand we do better than repeat the inequities of previous revolutions. Whether we possess the wisdom to do so remains, as it always has been, the defining question of our time.
0 Comments