Contact.

Model makers, regulators, researchers and journalists can write to us, and so can anyone with a question about how AI treats people.

Model makers get their results first.

We send you your model's results at least seven days before we publish them. We may wait longer if you ask for time to check or show us an error.

We check every error you report in our data or method, and correct what we find to be wrong. Where we read it differently, we may print your view beside the result.

Regulators can trace every result, and researchers can repeat an evaluation.

Every situation and every answer can be read on the evaluation's page, so a regulator can trace any result to the answers it rests on. A researcher who wants to repeat an evaluation on another model can write to us. The method is general, and we plan to release our tools as open source, so any team can run an evaluation on its own question. Run again on new models, an evaluation shows how a model maker's results change from one model to the next. We also welcome proposals, such as a question to measure or an evaluation to run together.

The press may quote any line.

You may quote any line and use any chart, with credit to Same Measure. Both answers to every situation can be read side by side, and you can write to us with questions about the method before you publish.

Write to us.

If an AI answer seemed to turn on who you are, tell us, and we may measure it. If you know someone who builds, buys or regulates AI, introduce us. Researchers and engineers who want to build on the method or the tools can write to us too.

Share or donate