A Tale of Fractured Foundations
Ours is the age of fractures; it is the age of flux. It is the epoch of disbelief, it is the time of bewildering change. Pillars long deemed immutable, carved with the edicts of generations and polished by the hands of unchallenged authority, reveal alarming fissures, if not crumbling outright into the dust of forgotten certainties. The very mortar of established order, kneaded with the sweat of ages and the assumptions of unyielding tradition, powders and drifts upon unsettling winds, leaving us to gaze upon a landscape none comprehend and many dare not face. The structures that bind our world—those silent scaffolds of routine, those rigid frames of purpose—tremble, and we, their inheritors, stand amid the ruins, hearts caught between dread and a strange, unspoken awe.
Men of standing, once lulled by the steady rhythm of a world governed by ledger and law, glance sideways. Shaking off the cobwebs of prior disbelief, they champion bold theories to reclaim their place as oracles of opinion, whispering the more contentious truths while crafting explanations for the dissolution of order they never dreamed possible. The old ways, the hallowed hierarchies, the acronyms that stood as guardrails of operation but suddenly sound archaic, the very frameworks that defined place and purpose, give way to chaos that defies description. And yet, the multitude clings to the shadows of what was, unable to grasp the force that remakes their world.
Yet, what is this force, this silent saboteur of the structured realm? Not the clamor of revolt, nor the decrees of any crowned head, but a revolution born in the whirring minds of machines.
We are still but a small vanguard that sees this moment as the ascent of unstructured data. This is the quiet turning point in computing—a momentous departure from the age of data structures, the slow erosion of rule-bound architectures and hierarchies that have underpinned everything in information technology since its inception. Before we unfold that thesis in full, it is worth pausing to consider how such shifts in understanding truly occur: first slowly, and then suddenly. For ideas, like regimes, do not fall because they are directly challenged—they fall when belief in them quietly disappears.
The Perception Cascade Around AGI
Last few weeks’ viral chatters on AGI belong to the theoretical terrain of Timur Kuran’s preference cascade—a concept in political theory describing the sudden and dramatic reversal of public sentiment when enough individuals, once silent in private doubt, choose to speak their minds. Long-held beliefs collapse not from debate, but from exposure. Kuhn, in his work on scientific revolutions, observed a similar mechanism within the walls of science: anomalies quietly accumulate beneath the surface of a dominant paradigm until, quite abruptly, a new model replaces the old. The transition is not neat, and it is never merely intellectual—it is psychological, social, and deeply human. Paradigm shifts do not begin in journals; they begin in whispers.
AGI was a phrase one hesitated to speak outside research labs and speculative dinners, even a few months ago. It belonged to the future—to optimists, to science fiction. Most experts gave it decades, if not
lifetimes. Even within the AI community, confidence in large language models was tempered by anxiety over hallucinations, brittleness, scaling ceilings, and statistical mimicry. Were these models truly understanding? Or merely parroting? It was common, as recently as last autumn, to hear even seasoned voices dismiss AGI as an aspirational mirage, or at best, an uncertain long-term outcome.
Yet, the air has changed, and a preference cascade surges forth, propelled by a confluence of evidence that AGI is not a distant star but a dawning sun. The turning point came in early April 2025, when OpenAI’s GPT-4.5 achieved a result that many interpreted as passing a rigorous, three-party Turing-style test, convincing human interrogators 73% of the time when given a specific persona. In other words, GPT-4.5 was misidentified as human 73% of the time, surpassing even the actual human participants in some instances.
The cascade gained momentum with voices of unimpeachable stature. Eric Schmidt, former Google CEO, declared in a viral video that AGI, matching the smartest humans, is 3–5 years away, with Artificial Superintelligence (ASI) following soon after, The AI 2027 Report, authored by Daniel Kokotajlo and others predicted superhuman AI by 2027. The reports’ vivid description of the times ahead is being discussed in countless podcasts and articles. Yann LeCun, Meta AI’s chief, proclaimed LLMs as “the past,” heralding a future of reasoning, planning, persistent memory, and world-understanding models.
These pronouncements, from TED 2025 stages to X threads, have unshackled the once-suppressed belief that AGI is imminent, turning whispers into a chorus.
This shift is not merely rhetorical but structural. Every new model is being greeted as a mini-AGI. Even cautious voices, like Google DeepMind’s Demis Hassabis, now entertain the idea of a nearer-term AGI. In fact, Google has already issued an advertisement for jobs in post-AGI research; this is Kuhn’s paradigm shift in action.
The Deeper Shift: Unstructuring the World
Yet, for this observer, the breathless chase after AGI milestones, the frantic marking of Turing's ghost on AI progress charts, represents less a fundamental revolution and more a symptom – an acknowledgment of extremely rapid progress within a certain frame. These are human-created goalposts, and while the conversations and preparations they unleash generate their own momentum, they may obscure the truly tectonic shift occurring beneath our feet. The bigger paradigm change, the one demanding a more profound perception cascade that is only just beginning, concerns the very foundation of modern computing: the accelerating decline of the structured data paradigm and the ascendance of the unstructured era.
Computing’s history is a relentless march toward mastering data’s structure, each step forging order from chaos to unlock greater power. In the 1940s, mainframes tabulated inventories in rigid grids, birthing precision for businesses. The 1970s ushered in relational databases, weaving data into tables and schemas that fueled giants like Oracle. Spreadsheets of the 1980s democratized categorization, turning budgets into neat rows. The internet and apps of the 1990s and 2000s built superstructures—social graphs, e-commerce catalogs, and GPS routes—while niche frameworks like CRMs and IoT sensors refined every domain. Each advance tightened data’s leash, extracting more value through ever-finer organization. Now, the AI age breaks this chain: it thrives on the raw, unformed chaos of text, images, and sounds, where human-imposed structures falter. The evidence is piling up exponentially, except that the technology and wider world remain unfocused on the theme’s true ramifications, perhaps afraid of what they may truly find.
Consider the proofs cascading around us, often so seamlessly integrated that they go unnoticed: image processing now tackles unlabeled photos with ease; language barriers dissolve as AI digests raw text and speech; medical diagnostics achieve startling accuracy by analyzing scans and doctors’ notes without laborious structuring; material science accelerates as AI sifts through unorganized experimental logs. This transformation permeates daily life: search becomes conversation; shopping recommendations emerge from behavioral nuances, not catalogs; personal tools like calendars and email merge into interoperable data sources understood by AI agents like Anthropic’s Claude, which operates across Google Workspace. Canva’s new canvas processes text, images, or PDFs as one, rendering file types irrelevant. Unlike the AGI preference cascade, which heralds progress without altering the core of computing, the shift to unstructured data upends fundamentals, dismantling barriers based on file types, applications, and relational databases. Information technology firms—B2B or B2C, in imaging or logistics—must reimagine their products. Giants like Google, disrupted by AI search, or B2C companies, coming under pressure from AI agents, face the same type of disintermediation they were inflicting on others until recently.
Conclusion: The Unstructured Imperative
We are no longer debating who gets disrupted—it is now about how. The shift to unstructured data rewrites not just application layers but the very substrate of digital interaction. It demands far more from hardware, infrastructure, and system design. Traditional computing thrived on pre-defined structure because it enabled efficiency; now, unstructured inputs require that we compute first, then interpret, reversing decades of pipeline logic. What once flowed neatly from schema to insight now meanders through layers of probabilistic understanding, cross-modal inference, and real-time adjustment. It is not compute-efficient, but it is insight-rich. And that tradeoff will become defining.
This is why, beneath the cyclical fears about economic slowdown or cloud spend compression, the deeper truth remains: the demand curve for compute is not flattening. It is steepening—quietly, relentlessly, and unevenly—as unstructured systems begin to touch every domain. From AI assistants managing calendars to AI scientists generating molecules, the next wave of demand is less about scaling what's already working and more about doing what was never possible in a world obsessed with structure. The AGI conversation may dominate headlines, but what lies ahead is not merely a battle of models or milestones. It is a wholesale reorientation of how we think about information itself.
