AI Societal Collapse: Pre-AGI Risks Exposed

The Unseen Precipice: Civilization’s Collapse Before the Dawn of AGI

By C. Rich 

As of December 1, 2025, the discourse surrounding artificial intelligence remains fixated on a singular, distant horizon: the arrival of artificial general intelligence (AGI) and its superintelligent successor, entities poised to either elevate or extinguish humanity. Experts like Roman Yampolskiy and Tristan Harris compel us to confront the existential perils of these technologies, uncontrollable optimization loops, misaligned goals, and a p(doom) that edges perilously close to certainty. Yet, in my view, this preoccupation obscures a more immediate and insidious threat. I contend that the incremental advancements of today’s narrow AI will precipitate a societal unraveling—economic, social, and geopolitical, long before AGI materializes. This pre-AGI fracture will not herald apocalypse through silicon overlords but through the quiet erosion of human systems, rendering the pursuit of godlike intelligence not merely unwise but impossible. The metrics of disruption, drawn from recent projections and analyses, paint a stark picture: a world buckling under its own accelerated transformation.

The economic dislocations alone suffice to destabilize the foundations of modern civilization. Projections for 2025–2027 forecast an unprecedented wave of job displacement, with the World Economic Forum estimating that AI will supplant 85 million roles globally by the end of this year, even as 97 million new positions emerge, a net gain that belies the turmoil of transition. Yet, this calculus overlooks the velocity of change. Goldman Sachs anticipates that AI adoption could displace 6–7% of the U.S. workforce within the next two years, targeting sectors from manufacturing to administrative tasks, where automation efficiencies render human labor obsolete. More alarmingly, a 2025 report warns of 83 million jobs worldwide being lost by 2027, encompassing both low-skilled manual labor and high-skilled cognitive roles in fields like software engineering and legal analysis.

Economist Anton Korinek, in a Harvard Business School analysis, projects that AGI precursors could drive wages toward zero by compressing labor’s marginal value, necessitating radical redistributive measures like a universal basic income to avert mass unemployment. Without such interventions, politically fraught in an era of fiscal austerity, these shifts will not foster abundance but ignite inequality on a Gilded Age scale. I envision supply chains fracturing as underemployed masses withdraw from participation, corporations hoarding AI-derived trillions while governments grapple with ballooning social safety nets. The result? Not a seamless pivot to post-labor economies, but widespread unrest: protests escalating into riots, as seen in early 2025 European manufacturing strikes, and a talent exodus from innovation hubs, starving the very R&D pipelines that propel AI forward.

Compounding this economic strain is the insidious corrosion of social cohesion, fueled by AI’s ability to spread misinformation. Deepfakes and generative tools, already deployable at scale, have infiltrated electoral processes with alarming subtlety. In the 2024 global elections, AI-fabricated content contributed to a 25% surge in voter distrust, manifesting not in overt scandals but in subtle erosions of shared reality. Looking to 2025, analyses from the Brennan Center for Justice predict amplified threats, with hyper-personalized propaganda, tailored via predictive algorithms, poised to undermine democratic integrity in battleground states. Bipartisan legislation, such as Congresswoman Julie Johnson’s bill introduced in September 2025, seeks to criminalize deceptive AI in campaigns, yet enforcement lags behind technological agility.

From my perspective, this is no mere digital nuisance; it is a solvent dissolving the glue of civil society. Polarization, already at fever pitch, will intensify as echo chambers harden into fortresses of fabricated consensus. Public discourse fractures into irreconcilable narratives, urban elites extolling AI’s efficiencies while rural communities decry cultural erasure, fomenting a crisis of legitimacy. Historians like Niall Ferguson warn of “megaslums” emerging from demographic stagnation and technological obsolescence, where disenfranchised populations, amplified by AI-orchestrated grievances, render governance untenable. In this milieu, trust evaporates, institutions falter, and the social contract unravels, diverting scarce resources from speculative AGI pursuits to mere survival.

Geopolitical fissures, meanwhile, will exacerbate these domestic tremors, transforming localized disruptions into global convulsions. AI’s infrastructural voracity, projected to consume energy equivalent to aviation by 2027, strains grids and ignites resource wars over rare earths and compute capacity. A Yale Insights forecast anticipates an “AI bubble burst” by late 2026, with annual investments surpassing $1 trillion, precipitating financial contagion reminiscent of 2008. Nations, locked in a multipolar arms race, will prioritize defensive AI—drones, cyber tools—over collaborative safety, as evidenced by escalating U.S.-China tensions over semiconductor exports. This zero-sum dynamic, echoed in X discussions where users decry the “lightspeed race” to AGI as self-destructive, risks proxy conflicts that hemorrhage capital and expertise. I see a cascade: blackouts in overtaxed data centers trigger economic shocks, misinformation-fueled nationalism erodes alliances, and talent pools—already thinning amid burnout and ethical defections—evaporate. The irony is profound: the very technologies accelerating toward AGI will, through these strains, sabotage their own ascent.

This pre-AGI implosion, in my estimation, constitutes a self-arresting mechanism, one that averts superintelligent cataclysm at the cost of civilizational regression. Metrics from PwC’s 2025 AI Jobs Barometer suggest that while AI enhances productivity in automatable roles, the societal ripple effects, unemployment spikes to 20–30% in vulnerable demographics, could halt innovation dead. Discussions on platforms like X amplify this foreboding: observers like Rohan Paul foresee wage collapse and political upheaval within 2–5 years, rendering AGI’s computational demands unfundable amid austerity. Far from the triumphant singularity, we risk a fragmented stasis: pockets of elite enclaves hoarding narrow AI for surveillance and sustenance, while the broader edifice crumbles into feudal disarray. This is not hyperbole but a logical extrapolation from the data, a world too fractured to sustain the trillion-dollar gambles on generality.

In reflecting on this trajectory, I urge a pivot from speculative dread to fortified resilience. Policymakers must enact international impact assessments, channeling AI’s bounty into universal services rather than unchecked scaling. Individuals, too, bear responsibility: cultivating analog communities, mastering AI as allies rather than adversaries, and amplifying calls for humane pauses, as Harris advocates. The precipice looms not in silicon transcendence but in our failure to steward the transition. By December 2025, with disruptions already manifesting, the question is not whether AGI arrives, but whether we endure long enough to face it.