Verified human respondents: why they matter for research and AI

A verified human respondent is a real, unique person whose identity has been checked, so one person cannot answer as many people and bots cannot pose as respondents. On Sibr, every Pi account counts as a verified human, because Pi Network only lets identity-verified people hold an account.

Sibr is a polling, survey and paid-study app on the Internet Computer, open to Pi Network users and everyone else. This guide explains why verification matters, how Sibr handles it, and where its limits are.

Why does it matter who answers a survey?

A survey's results describe the people who answered it. If some answers come from bots, or from one person running many accounts, the results describe nobody. The risk grows when answers are paid, because rewards give people a reason to cheat.

Problems that verification helps with:

• Duplicate people. One person with ten accounts counts ten times.
• Bots and scripts. Automated answers can fill a survey in minutes.
• Made-up profiles. Someone claims to be a nurse or a parent to qualify for a study.

Verification does not fix every problem. A real person can still answer carelessly. That is why Sibr pairs verification with quality checks, covered below and in How to get honest answers.

Why do AI teams need verified humans?

Teams that build AI models use human answers to train and test them: people rate model replies, compare two answers, or write examples. The point of that data is real human judgement. If the answers come from bots or from other AI tools, the data is misleading.

Two things make human data usable for AI work:

1. Real, unique people, so the answers reflect a range of humans.
2. Clear consent for that use, so people know their answers may train a model and agreed to it.

On Sibr, both are built in. Earn studies are for verified humans, and AI training is always a separate, unticked choice in the consent step. See Paid surveys with Pi.

How does Sibr verify people?

Sibr relies on Pi Network's identity checks rather than running its own.

Pi Network runs a Know Your Customer (KYC) process to check that its members are real individuals, keeps a strict one account per person policy, and requires KYC before members move to its mainnet. Sibr treats every account signed in with Pi as a verified human on that basis.

Internet Identity gives each app a private identifier and does not share your email address with Sibr. That is good for privacy, but it does not show that one person holds only one account, so those accounts can create and answer free polls and surveys but cannot earn.

What does Sibr learn about me from Pi?

Very little. When you sign in with Pi, Sibr keeps only your Pi user ID and Pi username. It does not store your access token and does not receive your real name, email address or wallet details. The identity check happens at Pi, not at Sibr.

What else protects study quality?

Verification is the first layer. Sibr adds others for paid studies:

• One response per account for every survey and study.
• Timing measured by the Sibr canister, so answers far faster than the estimated time are held for review.
• Attention checks, up to three per study, judged by the canister.
• Holds, not instant rejection. The creator reviews flagged answers and must give a reason to reject.
• Limits on starts: up to 3 studies in progress at once and up to 10 starts per hour per account.

What are the limits of verification?

Honesty matters here. Verification on Sibr proves a Pi account belongs to an identity-checked person. It does not prove:

• that the person answered carefully (quality checks help with that),
• what the person says about themselves in a profile, such as their job,
• where the person is. Country comes from the respondent's profile, and Sibr does not use IP addresses or device fingerprints for targeting.

How to do it on Sibr

As a respondent:

1. Sign in with Pi at sibr.online. Your profile shows your verification status and what it unlocks.
2. Open Earn to see paid studies for verified humans.

As a creator:

1. Sign in with Pi, which you need to fund a study.
2. In the survey builder, turn on Make this survey a paid study. Studies are listed in Earn for verified people.
3. Add screener questions, attention checks and a review window.
4. In your CSV export, each row shows the respondent's verification tier along with their answers, but never their username or account. See Results, insights and CSV export.

Sources

• Pi Network wiki, KYC: https://pinetwork.atlassian.net/wiki/spaces/WIKI/pages/26836993/KYC