John Locke

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI asks the reader to see the machine clearly: what it is, what it lacks, and what its existence changes about the human situation. Locke is the cycle's epistemologist of limits. He approaches the question of mind not by asserting the machine's humanity or denying its significance, but by identifying exactly which faculties are present and which are absent—a method of calibrated honesty that the AI discourse, which oscillates between worship and dismissal, desperately needs.

His blank-slate framework is the most precise available description of how a language model is made. But it also locates the machine's central limitation: a mind with only one of Locke's two fountains, sensation without reflection, is a mind that cannot notice its own confusion, cannot distinguish what it genuinely knows from what it is projecting, and cannot turn upon itself to ask what might be missing from the record it has absorbed. The companion question, which Locke presses on the reader of his Essay, applies with doubled force to the machine: what was left off the page, and how would you know?

His labor theory of property cuts directly to the most contested economic question the technology raises. The value concentrated in a trained model descends from the prior labor of every person whose writing filled the corpus—and on Locke's own premises, those people had a property in that work before any lab existed. His proviso—that appropriation from the commons is legitimate only where there is enough, and as good, left for others—provides a precise test that the displacement of the very workers whose labor trained the systems fails: the taking copies, but leaves the livelihood diminished. And his insistence that legitimate authority requires the genuine consent of those it governs names exactly what is absent in the deployment of AI systems over populations who never agreed.

Origin

Born in 1632 in Somerset to a country lawyer who fought on the Parliamentary side in the English Civil War, Locke trained as a physician at Oxford while absorbing the new natural philosophy of Boyle and Newton. His most enduring work, An Essay Concerning Human Understanding, was composed over nearly two decades—he began it in response to a discussion among friends about morality and religion that he thought had reached an impasse because no one had first asked how far the understanding could reach. The Essay appeared in 1689, the year after the Glorious Revolution in which Locke had conspired with William of Orange to displace James II, and the Two Treatises of Government, which provided the philosophical justification for the revolution, appeared the same year. Locke was simultaneously the period's most careful philosopher of knowledge and its most politically consequential theorist of legitimate authority; the combination is what makes him so productive for the AI moment, which is at once an epistemological problem and a political one.

He was a physician in temperament as well as training: empirical, precise, suspicious of untested speculation, committed to working only within the reach of the available evidence. The Essay opens with the image of the under-laborer, clearing the ground of rubbish so that more productive thinkers can build—not extending knowledge but mapping its limits honestly. This is the intellectual posture the AI discourse most requires, and it is the one Locke practiced most consistently. Where the dominant voices about artificial intelligence claim certainty—about what the systems are, what they will become, how far they understand—Locke's method insists on proportioning belief to evidence and conceding ignorance where ignorance is all we have.

Key Ideas

The blank slate and its two fountains. Locke held that the mind is at birth white paper, void of all characters, furnished entirely by experience—and that experience has two fountains: sensation (ideas of external things) and reflection (the mind's notice of its own operations). A language model is the most literal blank slate ever constructed: it begins as noise and acquires everything it has from its training corpus. But it has only one fountain. It lacks reflection in Locke's precise sense—the inward eye, the faculty by which a mind observes itself perceiving, doubting, believing—and its introspective talk is sensation-side data passed through, not the genuine first-personal acquaintance with its own operations that Locke made one of the pillars of the human mind.

Personal identity as continuity of consciousness. Locke pried personal identity loose from the soul and grounded it in consciousness: a person is whatever thinking being can consider itself as itself across time, appropriating past experience as its own through memory. The criterion is not substrate and not mere storage; it is appropriative, self-referring, continuous awareness from the inside. A model with a context window or external memory has storage; whether it has the consciousness that appropriates that storage as its own experience—that says of past outputs these were mine—is a separate question, and storage alone does not settle it. By Locke's criterion, the machine today is almost certainly not a person; by the same criterion, the question is not closed by its being a machine.

Labor, property, and the proviso. Locke grounded property in labor: when a person mixes their labor with what is common, the product becomes their own. Applied to AI training, the framework does not give the lab a clean title—it also establishes the prior claim of every laborer whose work fills the corpus. His proviso tightens the discipline: appropriation is legitimate only where there is enough, and as good, left for others. A model that learns from a writer's work and then competes that writer out of their market satisfies the proviso in letter (the original files remain) while violating it in substance (the livelihood is gone). Locke's labor theory does not resolve the ownership question; it supplies the exact vocabulary in which its difficulty becomes visible.

Consent and the limits of authority. Legitimate authority over a free person derives only from that person's consent, given for the preservation of their life, liberty, and property, and always conditional on the authority serving those ends. AI systems are deployed over populations who were not asked. The defender of the status quo will invoke tacit consent, but Locke's own conditions for tacit consent presuppose a genuine alternative—the possibility of withdrawal. When there is no exit from systems that saturate ordinary life, tacit consent is drained of the meaning that made it morally weighty. And when the authority turns against the ends for which consent was given, Locke's framework names the situation precisely: the governed retain the standing to refuse.

The limits of knowledge and the discipline of humility. Locke's deepest contribution is methodological: the Essay is an exercise in a mind measuring its own reach. Knowledge is the perception of the agreement or disagreement of ideas; everything else is probability, requiring calibrated confidence rather than assertion. A model that generates fluent confident assertions in the same register regardless of whether it has grounds commits Locke's cardinal epistemic sin at scale—the failure to distinguish knowledge from judgment, certainty from extrapolation. His discipline, transposed to AI output, means treating fluency as no evidence whatever of warrant, and demanding for any claim that matters the kind of independent grounding the model's confidence cannot supply.