Definitely would not be easy. First of all the mainstream llms are trained to be honest, and this requires lying convincingly. Second, this involves 8 hours of interviews with expert judges, one "claudism" could give it away.
All these problems can be fixed by a "pretend to be an average human to pass a turing test" prompt
All these problems can be fixed by a "pretend to be an average human to pass a turing test" prompt