This is where a lot of the misalignment problem in AI is going to come. We can build really useful, very smart machines - intelligent in terms of sheer knowledge, powerful in terms of speed, highly capable of selection and prediction; but it’s not going to have the full ability to *understand* and weigh things on the many levels that humans do. There’s just that “something” that can’t be built into an AI or prompted to develop there.
(One useful outcome may be that AI causes us to question the materialist bias in our science, and pursue some new ideas and possibilities. While scientists are generally less ‘believing’ than the general public, at least half of them do believe in some kind of higher or universal spirit or power.)
Are their Large Language Models (LLM) that have all of the Pravda archives as part of their "data"? Does the LLM programming state that communism is just another culture?
If a LLM has both AlGore's Inconvenient Truth and Michael Crichton's State of Fear as part of it's data, would it be able to determine that BOTH are fiction?