AI, Profit and the Public Interest
In my 2024 blog on digitalization and AI, I made the point that AI as the intelligence layer of the digital economy promises to be both a blessing and a curse for the global community. While no one knows how digitalization and AI specifically, will reshape societies — only that it will — my thinking at the time, encouraged by movements and agreements around the ethics of AI, was that the global community was aligned around the creation of a global digital economy underpinned by AI technology but constrained by actions, decisions and policies that serve the common good.
Two years on, I am not so sure. The uncomfortable truth is that the architecture of the digital economy and use of AI technology is not designed with the public interest as a primary constraint. It is designed to be profitable.
✦
01 The Profit Engine of the Machine
The technology industry tells a compelling story: that it is working to build a better future by democratising access to knowledge and, in doing so, lifting billions out of inefficiency and poverty. It is a story told by the wealthiest corporations in human history to — in my view — justify an unprecedented accumulation of power and profit.
On 1 June 2026, Anthropic announced that it had filed for an IPO, joining SpaceX and OpenAI in planning to list its shares. According to The Economist, the three potential IPOs could add as much as $4 trillion to the market value of listed American companies in a matter of months.
AI firms attracted $258.7 billion in global venture capital investment in 2025 alone — 61% of all VC globally, dwarfing the GDP of roughly three quarters of the world’s nations (OECD Analysis, February 2026). Gartner separately assessed total worldwide AI spending — including enterprise software, hardware and services — at nearly $1.5 trillion in 2025. Microsoft, Google, Meta, Amazon and a handful of other firms are spending at a pace and scale without historical precedent, with returns measured not just in dollars but in data, influence and market position.[1]
“They are locking in dependencies, capturing data flows, and extracting rent from the digital lives of billions of people and institutions. It’s less about building the future and more about buying it.” — Zuboff, 2019
Of course, the profit motive is not inherently sinister. Competition has driven genuine innovation and many AI applications[2] offer real social value. But the market logic of AI investment directs the most resources toward applications that maximise engagement, monetise attention and reduce labour costs — rather than toward problems where the returns are diffused or long-term.
✦
02 Digitalisation as Displacement
The economic disruption is already underway. Across customer service, legal drafting, medical coding, financial analysis, software development, content creation and logistics coordination, AI systems are absorbing tasks that previously required years of human training and delivered middle-class wages. The McKinsey Global Institute estimates that between 400 and 800 million workers globally may need to transition occupations by 2030 — not because their work disappears entirely, but because the nature of it changes faster than institutions can adapt.
“The gains from AI productivity tend to accrue to capital and to a narrow class of highly skilled workers; the costs tend to fall on those with the least power to absorb them.” — Acemğlu & Restrepo, 2022
The historical precedent — that technology creates as many jobs as it destroys — is true over long time horizons but offers little comfort to the radiologist whose diagnostic role is automated in the next five years, or the paralegal whose research function is replaced by a large language model at a fraction of the cost.
✦
03 An Architecture of Control
Digitalisation also changes power. The granular data flows generated by digital activity — who searches for what, who speaks to whom, where people move, what they buy and believe — create capacities for social monitoring and control that have no precedent in human history. In authoritarian contexts, these tools have enabled the suppression of dissent with extraordinary efficiency. But the concern is not limited to authoritarian states.
In liberal democracies, the same data architectures that enable personalised advertising also enable political micro-targeting, algorithmic content curation and the construction of information environments in which people are increasingly exposed only to views that confirm their existing beliefs. The fragmentation of shared reality — already visible in the politics of the United States, the United Kingdom, Brazil and beyond — is not an accidental by-product of social media. It is, to a considerable degree, the result of business models that reward engagement above accuracy, outrage above nuance, and certainty above complexity.
AI compounds these dynamics. Generative AI makes it cheaper and faster to produce persuasive disinformation at scale. Deepfake technology makes it increasingly difficult to distinguish authentic from fabricated media. Recommendation systems, now augmented by increasingly capable AI, grow ever more adept at identifying and exploiting individual psychological vulnerabilities. The societies best equipped to navigate these challenges are those with strong public institutions, high levels of media literacy and robust regulatory frameworks — precisely the capacities that decades of underfunded public sectors have eroded.
✦
04 In the Public Interest?
So, the question is not whether AI creates value but how that value is distributed and when. For developing nations — especially the small island states of the Caribbean, many of whom are still building out basic digital infrastructure — the risk is particularly acute. Arriving with the force of global capital behind it, AI is not waiting for the Caribbean to decide how it feels about it. I explore this issue with some depth in my next blog.
What is clear is that the digital economy is not a level playing field. It is one where first-movers, large markets and well-capitalised firms have compounding advantages. Countries that cannot afford to train their own AI models, build sovereign cloud infrastructure or retain technical talent will increasingly find themselves dependent on systems built elsewhere, for elsewhere, that they neither govern nor fully understand.
The architecture of the digital economy and use of AI is not designed with the public interest as its primary constraint. Does this mean public interest outcomes are impossible? No, but it does mean they require deliberate intervention.
A fact the Europeans understand well. The European Union’s AI Act — the most comprehensive regulatory framework yet attempted — establishes risk-based requirements for AI systems, with stringent obligations for high-risk applications in areas like employment, credit scoring and law enforcement. The Act is untested, and its enforcement will be contested, but it represents a serious attempt to subordinate AI development to democratic norms rather than simply to market logic. It may also serve as a model for CARICOM.
In the United States, regulatory capture remains a persistent risk. On 2 June 2026, the US President signed an executive order inviting vetting of top AI models for national security risks[3] — less than two weeks after postponing a similar policy over concerns it could stifle American companies’ lead in the global race with China. The signed order is framed explicitly around “America First cybersecurity” and “global AI dominance,” positioning AI governance primarily as a competitive geopolitical tool rather than a public interest instrument. The risks to workers, democratic integrity, privacy and climate are not in scope. In short, it is a tiny and strategically hedged step designed to be seen to act without meaningfully constraining the industry it purports to regulate.
In much of the Global South, there is simply insufficient regulatory capacity to engage meaningfully with the pace of AI deployment, leaving populations exposed to its effects with little institutional recourse.
✦
05 How AI Will Reshape Society — For Better and Worse
It would be incomplete and dishonest to discuss AI purely as a threat. The technology has genuine and substantial potential to address some of humanity’s most intractable problems. AI-accelerated drug discovery is already producing results in diseases that have resisted conventional research for decades. Climate modelling capabilities have been transformed. Agricultural AI is enabling yield improvements that could matter enormously for food security in a warming world. Educational tools have the potential — if equitably deployed — to give every child access to personalised tutoring previously available only to the affluent.
[4] As Mo Gawdat, former Google X Executive, put it:
“There’s absolutely nothing wrong with AI — there’s a lot wrong with the value set of humanity at the age of the rise of the machines.”
The issue is governance. The societies that will benefit most from digitalisation and AI are those that manage the transition deliberately: that invest in education and retraining, build genuine social safety nets, regulate the most harmful applications, maintain strong public institutions, and ensure that the productivity gains from AI are shared broadly rather than captured narrowly. These are political choices, not technical ones.
The societies most at risk are those that allow the transition to happen to them — that adopt AI systems built for other contexts, that lack the institutions to govern its effects, that allow market logic to determine outcomes that ought to be determined by democratic deliberation. For small and developing states in particular, the challenge is to build the agency and capacity to engage with AI on their own terms, rather than simply receiving it as a condition of participation in the global economy.
✦
Passengers or Co-Pilots?
Yes — the digitalisation and AI movement is substantially profit-driven. The dominant actors are corporations accountable primarily to shareholders, deploying technology at a scale and speed that outpaces democratic oversight. This means the distribution of its benefits and burdens is not being determined by what is most socially desirable — it is being determined by what is most commercially expedient. And, in the case of the United States, commercial expediency is further fuelled by the quest for leadership and dominance in AI advancements over China.
The shape of the society that AI will produce is not predetermined. Technology relies on the trajectory that politics and institutions give it. The critical question is whether democratic societies — and especially smaller, less powerful ones — can build the collective capacity to make meaningful choices about that trajectory, rather than simply adapting to choices made elsewhere. The benefits and burdens introduced by digitalization and AI require governments to make the right decisions for their people. Countries must decide: are they going to be passengers, or co-pilots?
✦
References
Acemğlu, D., & Restrepo, P. (2022). Tasks, automation, and the rise in US wage inequality. Econometrica, 90(5), 1973–2016.
The Economist. (1 June 2026). Can the stockmarket swallow Anthropic, SpaceX and OpenAI?
Crawford, K. (2021). Atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.
European Parliament. (2024). Regulation (EU) 2024/1689 — Artificial Intelligence Act.
Gartner. (2025). Worldwide AI spending to total $1.5 trillion in 2025 [Press release].
McKinsey Global Institute. (2023). The economic potential of generative AI: The next productivity frontier.
O’Brien, M. (2026). Trump signs executive order inviting vetting of top AI models for national security risks. Associated Press.
OECD. (2026). Venture capital investments in AI through 2025 [Policy brief].
OECD. (2024). OECD Digital Economy Outlook 2024.
Stanford HAI. (2025). AI Index Report 2025.
Tufekci, Z. (2018). YouTube, the great radicalizer. The New York Times.
UNCTAD. (2024). Technology and Innovation Report 2024: Catching technological waves — innovation with equity.
World Bank. (2024). World Development Report 2024: The middle income trap.
Zuboff, S. (2019). The Age of Surveillance Capitalism. PublicAffairs
[1]Of 195 countries only 50 have a GDP above this figure. For perspective, the combined GDP of all 14 CARICOM member states is roughly USD $120 billion.
[2]E.g. in medical diagnosis, climate modelling, agricultural optimisation.
[3]See: https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/
[4]Quoted in CNBC's coverage of his interview on the Diary of the CEO podcast with Steven Bartlett.
