Open science
Open Review Board
The MarcoFLY Framework claims to make an AI's answers more honest. That is a claim, and a claim has to be tested by someone other than the person who made it. This page shows how that test is going — including when it goes badly.
Towards 1000 independent validations
The result
There is not enough data yet.
The comparison between ratings with and without the framework will appear here once 30 reviewed validations are in. Below that threshold a difference between two averages is noise, and showing it would suggest a result that does not exist. We prefer a page that says “not yet” to one that appears to say something.
How they are distributed
By question type
- Creativity1
By language
- IT1
Who validated
No names yet — and the space is open. Whoever validates can choose to appear here or stay anonymous: it is a checkbox separate from the open-data consent, and it defaults to no.