The idea of artificial general intelligence (AGI), or machines capable of performing the full breadth of human cognitive tasks, has shifted rapidly from speculative fiction to a plausible technological horizon. This transition has been driven by a convergence of breakthroughs in large-scale machine learning, particularly the rise of transformer-based architectures, models such as GPT-5 and beyond, advances in multimodal systems, and the scaling of compute and data that has enabled increasingly generalizable capabilities across domains. Recent speculation that systems such as Anthropic’s Mythos Preview have supposedly demonstrated the ability to perform complex reasoning tasks without being explicitly trained to do so, shows the shrinking gap between narrow artificial intelligence (AI) and more generalized AI.
If AGI emerges, it could become the most transformative force in economic and social life since the Industrial Revolution. Its promise is extraordinary: near-total automation of intellectual labour, accelerated scientific discovery, rapid innovation cycles and an economic order in which scarcity begins to recede. For the first time in history, humanity may possess systems capable of generating the material foundations of prosperity without continuous human input.
Yet the promise of abundance is inseparable from more troubling questions: Who will own AGI’s capabilities, who will capture the immense value it creates and who will decide how that value is distributed? Technology alone does not deliver justice or equality; it magnifies the institutional structures into which it is embedded. If the world is genuinely moving toward AGI, societies must confront a challenge deeper than technological readiness: the redesign of the social contract among states, citizens and the private corporations that build and operate these systems.
At the centre of this debate lies a fundamental economic rupture. For centuries, labour has been the primary driver of productivity and the main channel through which individuals participate in economic life. Even in highly automated economies, such as Singapore, South Korea, Japan and Germany, humans remain essential to design, governance and innovation. AGI alters this relationship. A sufficiently capable system could autonomously design drugs, optimize supply chains, negotiate contracts, forecast markets and coordinate scientific research — tasks that today require large teams of highly skilled workers. If this occurs, economic output could rise dramatically even as traditional pathways to earning income erode. Productivity would no longer depend on human effort but on access to AGI systems, data and computational infrastructure. Those who control these assets — primarily large technology firms and a small number of states — would capture disproportionate returns, potentially concentrating wealth more intensely than during the platform economy.
This leads to the central dilemma: If AGI generates the wealth, who pays people? When labour is no longer the foundation of production, distribution becomes a political choice rather than a market outcome. One vision imagines states stepping in to redistribute AGI-generated value through universal basic income, social dividends or public wealth funds financed by taxes on compute, automation rents or AGI profits. In principle, this could enable a post-scarcity welfare state that guarantees economic security independent of employment. In practice, its feasibility depends on political legitimacy, administrative capacity and global inequality. Wealthy states may sustain such systems, while poorer countries risk deeper dependence on external AGI providers.
A contrasting vision places private corporations at the centre of distribution. If firms own the AGI systems that automate most economic activity, they may provide stipends, dividends or services to users, effectively becoming private welfare institutions. Economic efficiency may improve, but democratic accountability would weaken. Citizenship would matter less than platform membership, and the legitimacy of the state itself could erode as corporations assume responsibility for livelihoods.
A Future of Mutual Dependence
More plausibly, a hybrid model may emerge, in which governments and corporations negotiate shared authority over AGI development and distribution. This outcome is likely not by design but by necessity. Most states lack the technical capacity, compute infrastructure and pace of innovation to independently develop frontier AGI systems, while private firms lack the legitimacy, jurisdictional authority and risk-bearing mandate required to govern a technology with systemic societal implications. As a result, mutual dependence would push both actors toward co-governance. Early signals of this model are already visible. Governments are increasingly partnering with frontier AI firms through public compute initiatives, regulatory sandboxes and safety commitments, while companies engage proactively with regulators on issues such as model evaluation, alignment and risk mitigation. Proposals for “AI safety institutes” in the United States and the United Kingdom, public-private collaborations on frontier model testing and emerging discussions around global AI governance frameworks all point toward a negotiated model of shared oversight rather than unilateral control.
In such a scenario, AGI would function as critical infrastructure, privately built but publicly regulated, similar to energy, telecommunications or aerospace, though far more consequential. This would require entirely new legal and institutional architectures: frameworks for access, benefit sharing, safety oversight and cross-border governance. While these structures remain underdeveloped, their contours are beginning to take shape through ongoing policy debates, industry commitments and multilateral discussions — suggesting that the governance of AGI will likely emerge through incremental compacts rather than a single, fully formed institutional design. Layered onto these domestic challenges are powerful geopolitical dynamics. AGI’s benefits will not be evenly distributed.
Countries with advanced semiconductor ecosystems, large data sets, deep AI talent pools and massive compute capacity, such as the United States, China, India and parts of Europe, will dominate early development. Many others will become dependent on foreign AGI systems for health care, research, industrial optimization and even national security. This risks creating a new global divide not merely of income but also of intelligence access. Abundance may exist, but unevenly, thus becoming a tool of geopolitical leverage akin to oil in the twentieth century, though far more pervasive.
Beyond economics and geopolitics lies a deeper philosophical question: What constitutes a good life when work is no longer central to human identity? Employment has long provided meaning, structure and social belonging. Hannah Arendt, for instance, situated labour and work at the centre of the human condition, linking them to identity and participation in public life, while Max Weber described work as a moral and social calling embedded within modern economic systems. If AGI renders most labour unnecessary, societies must cultivate alternative sources of value, such as through creative pursuits, civic engagement, education, care or new forms of contribution. Abundance could enable a renaissance of human flourishing, or it could give way to alienation and purposelessness if institutions fail to adapt.
The legal implications are equally profound. AGI challenges core doctrines of property, liability and rights: Who owns the knowledge autonomously generated by machines? Who is responsible when AGI systems cause harm? Should access to AGI become a legal right if it determines opportunity and survival? Should data be treated as labour, property or a collective resource? The answers will shape whether AGI becomes a democratic instrument or a new form of feudal power.
Finally, abundance is not guaranteed. AGI could just as plausibly produce mass unemployment, political instability and authoritarian control. It could intensify surveillance, accelerate arms races, or provoke conflict over computing resources and supply chains. In short, abundance is only one possible outcome among many.
The future AGI ushers in will depend less on algorithms than on institutions, laws and political choices. Technology may redefine what is possible, but governance will determine what is just. If societies act deliberately, building inclusive frameworks and sharing benefits widely, AGI could underpin the most prosperous era in human history. If not, it may amplify inequality, destabilize democracies and concentrate power beyond public control.
