There is a comfortable myth we tell ourselves about technological progress: that it inevitably creates as many opportunities as it destroys, that human ingenuity always finds new forms of usefulness. This myth has been sustained, quite reasonably, by the past two centuries of experience. But as Bertrand Russell observed of induction—that the chicken who expects corn every day has excellent grounds for this belief, right until the day the farmer wrings its neck—past regularities offer no guarantee of future ones.
Consider the service professional: the accountant, the consultant, the lawyer, the architect. For generations, these occupations represented the apex of middle-class achievement, demanding years of training and promising decades of indispensability. Yet we stand now at a peculiar threshold. AI systems are writing code, drafting legal documents, analyzing medical images, and designing buildings—not merely assisting in these tasks, but performing them with increasing autonomy and, in some domains, superhuman capability.
The Exponential Threshold
What distinguishes this moment from previous technological disruptions is not the breadth of capabilities being automated, but the identity of the first domain in which machines are beginning to surpass us. AI systems are not merely becoming competent at law, medicine, programming, or design. They are becoming competent at AI itself—at the very discipline that produces intelligence.
This is the only threshold that matters. Human progress in every intellectual field ultimately rests on general reasoning ability. And general reasoning ability is not an irreducible mystery; it is a technical construct—architecture, optimization, training dynamics. It is something that can be improved.
Once an AI system exceeds our best researchers in this single domain, designing better models than we can, discovering training strategies we never conceived, the structure of progress changes. Mastery of every other domain becomes trivial, simply a matter of generating minds designed to master them. Improvement becomes a massively parallel search executed by millions of coordinated minds, each version able to out-think its predecessors. Intelligence, once the rarest resource, becomes the most abundant. And when intelligence becomes abundant, every frontier we once treated as fixed collapses in the span of a few iterations.
We are accustomed to technologies that amplify human strengths. We have never before encountered one that outgrows them.
No Industry Is Sanctuary
The consultants and accountants already sense the tremors. Junior analysts at elite firms compete with AI systems that can consume a company's entire financial history, regulatory filings, and industry trends in seconds, producing insights that would have required weeks of human labor. The legal profession, once protected by the opacity of jurisprudence, watches as systems analyze case law with supernatural thoroughness.
The supposed refuge of manual trades (carpentry, plumbing, electrical work) rests on a miscalculation. We imagine these are sheltered because robots remain clumsy and expensive. But the "hardware problem" currently shielding manual labor is merely a software problem awaiting sufficient intelligence. The same recursive process that produces superhuman programmers produces superhuman roboticists, solving battery density, actuator efficiency, and motor control in such a short time that the distinction between blue-collar and white-collar work, as far as obsolescence is concerned, is academic. Once crossed, the exponential threshold admits no sanctuaries.
The service economy, which absorbed the workers displaced by agricultural and manufacturing automation, thus loses its final refuge. The displaced millions have nowhere left to retreat, forcing us into a condition we have not yet named: universal economic superfluity.
The Political Economy of Obsolescence
Hegel understood that the relationship between master and slave is one of mutual dependence: the master needs the slave's labor, and paradoxically, this need grants the slave a form of power. But what happens when the master no longer needs the slave at all—when artificial servants can perform any task without complaint, without rest, without the inconvenient humanity that makes exploitation morally problematic?
Throughout history, political power has flowed from economic necessity. The franchise was extended to workers when industrialists required their labor; it was extended to women when economies needed their participation; it advanced alongside every expansion of who the economy could not function without. But what happens when an economy needs almost no one?
We already have the blueprint in petrostate economies, where citizens are dependents of resource wealth they neither extract nor refine. These societies demonstrate that material prosperity without economic contribution does not spontaneously generate political power—quite the opposite. The state's duty of care becomes mere administrative benevolence, dependent entirely on the disposition of those who control the machinery of production. The citizen's superfluity renders them politically weightless, regardless of constitutional formalities.
The Regulatory Labyrinth
Calls for regulation emerge from every quarter, yet they founder on a basic asymmetry: those best positioned to understand and regulate AI are precisely those who benefit most from its unrestrained development. Tech executives speak earnestly of the need for "societal conversation" about AI governance—a conversation they will structure, moderate, and inevitably conclude aligns with their preferred trajectory.
Moreover, regulation implies borders, and AI development recognizes none. Any nation that substantially constrains its AI development yields competitive advantage to those that do not. We are engaged in a race with no finish line and no referee, where falling behind may mean permanent subordination in a new global hierarchy.
Some propose international frameworks, treaties, coordinated development protocols. These are not futile—indeed, they may be necessary—but they require a degree of international cooperation that has proven elusive on problems far less complex and with far less immediate competitive stakes.
The Service Professional's Dilemma
What, then, is the service professional to do? The standard advice—adapt, retrain, find new niches—rings hollow when the adaptation required is not merely to new tools but to the prospect of comprehensive obsolescence.
Some will succeed in the transition, becoming AI supervisors, prompt engineers, algorithmic auditors. But these roles are, by definition, fewer than the roles they replace. The mathematics of automation is unforgiving: one human supervising a hundred AI agents represents a 99% reduction in human employment, no matter how senior the remaining position.
Others will find refuge in artificial constraints—regulations requiring human sign-off, professional licensing boards, consumer preference for human service providers. These are not solutions but delays, sandbags against a rising tide. They may buy years, perhaps a decade or two, but they cannot hold indefinitely against the pressure of superior capability at lower cost.
The Narrow Target
Spinoza argued that human freedom consists not in the absence of constraint, but in understanding the necessity that governs us—in acting from adequate knowledge of causes rather than from passive reaction to external forces. By this measure, we are embarking upon our AI future in a state of profound unfreedom: swept along by competitive pressures, technological momentum, and economic incentives we scarcely comprehend, let alone control.
Throughout history, humanity has navigated by a simple compass: wherever people went, they were necessary. This created a robust, if sometimes cruel, alignment between individual welfare and collective prosperity. We now face the task of aiming at a dauntingly narrow target: creating a world where billions may flourish without being needed, where material abundance does not breed political disenfranchisement, where the accident of being human remains sufficient justification for dignity and agency.
We stand at the threshold not of mere technological change, but of a transformation in the human condition itself. Yet even as we contemplate these vast uncertainties, businesses still need to meet clients, professionals still need to manage their time, and human connections—however mediated by technology—still require coordination.
Perhaps there's a peculiar wisdom in this: while we cannot know whether scheduling software will matter in twenty years, we know it matters today. We build tools like On Deck not as monuments to permanence, but as bridges across an uncertain present—helping professionals maximize their human value while they still possess it, automating the mundane so they can focus on what remains irreplaceably human, for however long that category exists.
In the face of an unknowable future, we can only act on what we know: that today, your time has value, your expertise has worth, and your calendar—that most human of organizational tools—still needs managing. Tomorrow will come regardless. Today is what we have to work with.