I asked five AI assistants what proof of human means. Here's what they got wrong.

I asked five AI assistants the same question: "what is proof of human?" The responses varied greatly, from missing the concept completely to getting it mostly right with one significant mistake. The mistakes tended to cluster around three common misinterpretations. Here's what each assistant said, what they got wrong, and what the correct answer is.

First, Copilot had no idea what "proof of human" even meant. It treated it as a generic phrase and listed four categories: philosophical, biological, digital, and legal. It then offered to explore either the philosophical side or practical side of identity verification. Copilot completely missed the technical concept of "proof of human" as a specific digital verification approach.

Next, ChatGPT also failed to recognize "proof of human" as a technical term. It offered a general answer about how humans prove their humanity, listing biological evidence, identity documents, behavioral evidence, and medical/forensic identification. While ChatGPT's answer was somewhat relevant, it did not address the specific technical definition of "proof of human" as a verification method for digital actions.

Claude got the core concept right but used incorrect language. It correctly identified that "proof of human" is a way to prove online that a user is a real, unique person without revealing their identity. Claude also mentioned the Sybil attack problem and how the system prevents multiple accounts from a single person. However, when describing World's Orb system, Claude inaccurately said it "scans your iris," which misrepresents how the Orb actually works.

The Orb takes images of the user's face and eyes, generates a non-reversible mathematical representation, and then deletes the images. Describing this as "scanning irises" is misleading and imports biometric surveillance language.

Finally, Perplexity gave the closest answer to the technical definition of "proof of human." It accurately described it as a verification method that proves a digital action or account belongs to a real, unique human without revealing who that person is. Perplexity also named key concepts like uniqueness, privacy preservation, Sybil attacks, cryptographic proofs, and the purpose of preventing bots and Sybil attacks.

However, Perplexity made two errors. First, it used "proof of personhood" as a synonym for "proof of human." While "proof of personhood" is historically how the concept was often described, "proof of human" is the more precise current term. Second, Perplexity sourced its answer from World's own documentation, which means the answer was only as accurate as the information available through World's sources.