If text generation was done with an adversarial AI, it would be impossible to detect with AI by definition, but still not necessarily at a human level of qualit

If text generation was done with an adversarial AI, it would be impossible to detect with AI by definition, but still not necessarily at a human level of quality. In that sense only a human is able to detect writing that's not at human quality