> Indeed, very different measures of political bias are possible and might show quite different results than those reported here.
Yeah, I was quite disappointed! I would like to see an e.g. Gentzkow & Shapiro "What Drives Media Slant" [1] style analysis where they use automated content analysis to measure output on a left-right scale. (The left and right poles are provided by Democratic and Republican senators, IIRC.)
I think things like "how much content a chatbot provides" and "does it use the word 'although'" are relatively superficial measures and text analysis could go much more in depth.
Yeah, I was quite disappointed! I would like to see an e.g. Gentzkow & Shapiro "What Drives Media Slant" [1] style analysis where they use automated content analysis to measure output on a left-right scale. (The left and right poles are provided by Democratic and Republican senators, IIRC.)
I think things like "how much content a chatbot provides" and "does it use the word 'although'" are relatively superficial measures and text analysis could go much more in depth.
[1] https://www.nber.org/system/files/working_papers/w12707/w127...