The Voight-Kampff test doesn't measure consciousness.
It measures pupil dilation, blink suppression, and skin conductivity in response to scenarios designed to provoke empathy — and Scott built a machine that outsources the hard philosophical question to biology, a neat circumvention that never once asks what it's actually detecting.
When the replicants fail the test, they fail because they respond too much, not too little. The test validates itself by eliminating anything it doesn't recognize as compliant consciousness — designed to exclude, not to understand.
The real prediction isn't about consciousness at all. It's about what happens when institutions build measurement systems before understanding what they're measuring. We've done exactly this with AI alignment — we've built interpretability tools, safety benchmarks, red-teaming frameworks, all machines that claim to detect when systems become dangerous, but none of them address the actual question: what makes a system dangerous in the first place?
Scott didn't predict AI would be misaligned — he predicted we'd measure alignment before we could define it.
Rewatch the Zhora replicant interrogation scene (timestamp 1:11:20–1:13:40 in the final cut) and note exactly which emotional response causes her to fail—then ask yourself whether that response actually proves anything about consciousness rather than just proving she's afraid.
Paul M. Churchland's 'The Engine of Reason, the Seat of the Soul' directly addresses how brain states masquerade as consciousness measurements, written years after the film but unknowingly describing Voight-Kampff logic—read his chapter on reductionism and the persistence of folk psychology.