Possibility 1: It does mean the designers in which case some machines may get promoted to AI shortly after their designers die.
Possibility 2: It means anyone with access to the design in which case if the AI has capabilities beyond ours then one 'AI' may be able to explain the creative process of another in sufficient detail to preclude any of them being classed as AI. This test appears to assume human limitations. While we may one day understand enough neuroscience to explain human creativity in a hand-wavy fashion I doubt we'll ever get to the point where a human can explain another human's creative process in detail.
Another thought. Would adding a genuine random element make this test rather easy to pass? While a designer might be able to provide an explanation for a given creative work they could never guarantee it was the correct explanation. If the requirement is that nobody can provide a plausible explanation for the creative work then humans might not pass the test.