[quails8mydog]

I agree that it's an interesting question, that's why I spent so much of my free time reading it.

I'd also agree that using AI for sentiment analysis could be a good approach, I'm not an expert in the area, but I believe this is one of the things AI is best at. But it needs an extra step to translate that into bias. Establishing a sympathy baseline is my initial idea, but I haven't tested it and maybe there's something better.

Whether something is biased is less about how any given individual(s) feel about what's been said and more about if the different viewpoints are presented honestly. Though it can get really difficult to identify except in the most extreme cases. As you say, it's not just what's said where the bias occurs, but also in the choice of what not to say.