I think the Turing Test is usually described as an examiner is freely conversing with two subjects. The examiner knows that one subject is human and the other i

I think the Turing Test is usually described as an examiner is freely conversing with two subjects. The examiner knows that one subject is human and the other is AI. They are supposed to determine which is human and which is AI. If they can't achieve higher than 50% accuracy then the AI is deemed to have passed the Turing test. It would be a much easier version of the test if the AI only has to fool the examiner in a limited set of circumstances such as a scammer trying to qualify people as potential scam targets with a minimum of effort.