@DickForgetsHeRunsServices@paypig.org "someone play devil's advocate with me" It depends on how the AI is implemented. If it's implemented according to a fixed set of rules then the merit of having an AI editor over a human editor is that there will at least be consistency with the AI and a lack of political bias. If the fake news is "fake" it will tagged as fake regardless if it's left-wing or right-wing. devil's advocate to the devil's advocate: The most advanced AI capable of this is mostly implemented using machine learning via human-fed inputs. ie the AI learns what is fake news by being fed what humans have determined is fake news. It learns the political biases along with it. In reality there are no "inconsistencies" to the applications of a rule, simply other rules unwritten. These patterns are learnable by an AI. Even if the AI was somehow programmed on following a "principle of truth"- rather than a "pattern of truth" it still just will be matching up user claims against those on "fact checking" sites. At some point along the line there is a human input with the biases that come with that- all the AI is doing is automating the rest of the process. Even if it only looked at research papers and collected data rather than secondary conclusions about said data- there's still sampling bias and other statistical distortions (including outright lies) that plague what is ultimately a human-implemented process. devil's advocate to the devil's advocate to the devil's advocate: Then make the AI do the primary source collecting too. Robocop bouta btfo even more huffpost journalists.