Synthetic Intellects and Forged Laborers

In the [YOU] on AI Field Guide

The cycle treats AI as an amplifier: it carries whatever signal is fed into it, with terrifying fidelity, at unprecedented scale. Kaplan’s taxonomy gives that amplification a labor-market shape. The amplifier is, in his terms, a synthetic intellect of extraordinary power—a large language model that performs cognitive labor without any of the cognition we once assumed such labor required. The twenty-fold productivity multiplier measured in Trivandrum is the amplifier doing what amplifiers do to synthetic-intellect tasks: removing the friction between intention and output, so that one person can produce what previously required many. The engineer who felt like a master calligrapher watching the printing press arrive was experiencing, in personal form, the displacement that Kaplan’s taxonomy predicts at civilizational scale.

The taxonomy also illuminates what the cycle calls ascending friction—the thesis that AI does not eliminate difficulty but relocates it to a higher cognitive floor. In Kaplan’s framework, synthetic intellects are the mechanism of relocation: they absorb the specifiable, the pattern-rich, the mechanically executable, and push the remaining human premium toward judgment, taste, and the capacity to decide what should exist. The forged laborers, on their slower timeline, are working the same relocation at the level of physical labor, demanding that workers develop the specifically human capacities that the machine’s sensors and actuators cannot replicate.

Origin

Kaplan coined the terms in Humans Need Not Apply as a deliberate intervention against the dominant vocabulary of the field. The word intelligence—as he analyzed in later chapters of the same book—is an honorific we bestow on whatever cognitive feats currently impress us, and it imports a mythology of mind and consciousness that the underlying technology does not support. By replacing intelligence with synthetic intellect and robot with forged laborer, he focused attention on economic function rather than cognitive status, on what these systems do in the labor market rather than on what they might or might not experience. The reframing was both conceptual and political: it made visible a division that the unified vocabulary of “AI” had been obscuring, and it directed the policy conversation toward the two different disruption timelines that the two classes produce.

The taxonomy also drew on Kaplan’s direct experience building both classes. Teknowledge had built systems that analyzed and advised—proto-synthetic intellects. GO Corporation had attempted to build a device that combined intelligence with physical interaction—a proto-forged laborer. Having watched both from the inside, Kaplan understood that the obstacles were different in kind, not just degree, and that conflating them produced both bad predictions and bad policy.

Key Ideas

Capability, not consciousness. The taxonomy deliberately organizes machines around what they can do economically rather than what they might or might not experience. A synthetic intellect that passes professional exams without understanding anything is still a synthetic intellect; the question of its consciousness is irrelevant to its labor-market impact. This reframing cuts through the philosophical muddle that prevents most AI discussions from reaching the economic conclusions they need to reach.

Two timelines, two responses. Synthetic intellects advance at software speed and face no friction from the physical world. Forged laborers advance on the slower timeline imposed by physics, materials science, and the sheer difficulty of replicating the dexterity and adaptability of a human body. The two timelines demand different institutional responses: white-collar displacement is already happening at scale and requires immediate action; blue-collar displacement is coming but allows somewhat more time to prepare. Conflating the timelines produces either panic or complacency—neither appropriate.

The breadth of vulnerability. Previous automation debates assumed the dividing line between safe and unsafe work ran between manual and intellectual, routine and creative. Kaplan’s taxonomy revealed a different dividing line: between predictable and unpredictable. Both synthetic intellects and forged laborers are expanding the domain of the predictable. The radiologist reading scans and the truck driver navigating highways are, from the machine’s point of view, doing similar things: mapping patterns of input to appropriate output. Substitution turns out to threaten far more of the labor market than the comfort zone of previous generations’ analysis could accommodate.

The ownership implication. Once the taxonomy is in place, the implication is stark: when both cognitive and physical labor can be substituted at scale, the only reliable economic position is ownership. The person who owns a meaningful stake in the productive machinery of the economy will prosper as that machinery grows more capable; the person who owns nothing but their own labor will struggle no matter how hard they work. This is why Kaplan’s taxonomy leads directly to his proposals for financing retraining and spreading the ownership of capital—not as charity but as the structural response to a structural problem.