Wow! 80% of this post was mostly foreign to me, yet I deeply enjoyed being mystified. I also love that you call yourself a data scientist hobbyist :) When it comes to AI evaluating media or public policy, what do you envision? Also, what do you think would be the greatest realistic risk of AI becoming more sophisticated? No need to respond in depth if these are too big to answer via a Hive comment!
AI can read media reports and detect biases/misleading information, with the potential to make propaganda more transparent. AI could serve policy by reading a million little environmental signals for a variety of reasons, from understanding risk in a given situation to recommending interest rate adjustments.
The greatest near-term risk of sophisticated AI in my opinion is that it's likely to be used by the powerful to further consolidate their power. Good AI could already make a UFO hoax look believable.