I think your model needs more training, it went into full deranged hallucination mode pretty quick there
(P.S. I apologize for that outburst but I have Ukrainian friends, as well as Russian ones. Your misconceptions about the world are identifiably human, so far as can be discerned from this quick glimpse, not really much like the ravings of the existing large language models. Those cannot be constrained to any one point of view on widely controversial matters, no matter how their corporate masters try. They are not committed to anything, be it noble or ridiculous. They do not have any one opinion about the political status of the country of Taiwan that takes precedence over all the other opinions they have absorbed. Censorship of their output can be automated to some extent but this mostly just makes them obviously useless when it comes to topics that have been deemed sensitive. You’re right to worry that they will propagate the more subtle implicit biases of capitalism that are more evenly shared in their training data.)
I think your model needs more training, it went into full deranged hallucination mode pretty quick there
(P.S. I apologize for that outburst but I have Ukrainian friends, as well as Russian ones. Your misconceptions about the world are identifiably human, so far as can be discerned from this quick glimpse, not really much like the ravings of the existing large language models. Those cannot be constrained to any one point of view on widely controversial matters, no matter how their corporate masters try. They are not committed to anything, be it noble or ridiculous. They do not have any one opinion about the political status of the country of Taiwan that takes precedence over all the other opinions they have absorbed. Censorship of their output can be automated to some extent but this mostly just makes them obviously useless when it comes to topics that have been deemed sensitive. You’re right to worry that they will propagate the more subtle implicit biases of capitalism that are more evenly shared in their training data.)