Does it have a latent personality? How would it answer the questions on a 5-factor personality test? Would its results on the test be consistent with its behavior (generated text) in other situations?
Or, can it (like humans?) adapt its responses to suit the style of the question? Like if you start asking it lots of antagonizing questions, will it become more or less antagonistic itself?
