I don't see any way to define it in an easily verifiable way.

Pretty much any test you could devise, others will be able to point out ways that it's inadequate or doesn't capture aspects of human intelligence.

So I think it all just comes down to who is on the panel.