Synthetic Identity Probability Calculator

Treat the widget below the way you would a magic-8-ball: interesting to shake, never a basis for a real decision. Its value is in what it teaches about the limits of surface-signal detection — move the sliders and notice how easily the same signals that flag some AI text would also flag a careful human writer.

Why real detection is harder than this toy

The four sliders here — linguistic staleness, response cadence, contextual drift, semantic range — point at real signals researchers have studied, but collapsing them into a single number is exactly the oversimplification that makes consumer "AI detectors" unreliable. Modern language models vary their pacing, avoid obvious repetition, and hold context well enough that these surface tells fire inconsistently. And the deeper problem is symmetry: the same signals that flag some AI text also flag plenty of human writing. Non-native English speakers, people writing in a careful or formal register, and heavily edited prose all get caught by real detectors at uncomfortable rates, which is why several universities have quietly disabled AI-text detection rather than keep accusing their own students. A slider gadget like this one has none of the sophistication of those tools and all of the same blind spots, magnified.

What actually works

The durable answer to "is this real?" is not detection but provenance — cryptographic Content Credentials attached at creation, which our field guide explains in full. Where provenance is absent, the honest move is corroboration: reverse-image search, source history, reading laterally across independent sources, and weighing the account's behaviour over the artifact itself. None of that produces a tidy percentage, and that is the lesson this toy is really here to teach — anyone who hands you a confident number about a real person is selling certainty that does not exist. Treat this calculator as a demonstration of that fact, not an exception to it.

How to use this responsibly

There is exactly one responsible use for this widget: as a teaching aid to feel, first-hand, why surface-signal detection fails. Move the sliders while picturing a specific human writer — a meticulous non-native English speaker, say, or a lawyer trained to write in clean, structured prose — and watch the toy climb toward "synthetic" for writing that is entirely human. That experience is the lesson. It is the same failure mode that has led real institutions to stop trusting commercial AI-text detectors after they falsely accused real people. If you came here hoping for a quick way to check whether someone or something is "AI," the honest and more useful answer is on our field guide: look for provenance, corroborate across sources, and accept that a confident percentage about a real person is a fiction no gadget can honestly provide.

The bigger point

A calculator that admits it cannot do its stated job might seem pointless, but the admission is the point. The market is full of tools that make the same promise this one deliberately breaks — a clean number telling you whether a person, essay, or image is "real" — and they sell that certainty precisely because people want it badly enough not to check. Wanting it does not make it available. The most useful thing a synthetic-identity site can hand you is not a detector but a calibrated distrust of detectors, paired with the habits that actually work: verifying provenance, corroborating across sources, and judging behaviour over surface. Keep those, discard the gadget, and you are already harder to fool than the tools that charge for the privilege.