(and other discussion platforms)
I think a good use of AI would be to to highlight logical fallacies.
Like Community Notes, this could be an add-on to a comment. It would detail the fallacies contained in the comment, why they are fallacies and why they do not productively add to the dialogue.
I would also suggest, as a general rule of discussion, the first to use a fallacy loses. Or, score could be kept by keeping tabs on good points and fallacies.
A popular fallacy I’ve seen on Twitter of late is a combination of two fallacies: a personal attack (ad hominem) and changing the subject (red herring). It goes something like, “You appear to have a reading comprehension problem. I must be talking to an idiot. Have a nice day.”
I call this a combo fallacy because it’s a personal attack, but also designed to change the subject from the topic at hand to whether the person has a reading comprehension problem and to end the discussion with the user of it feeling they have won.
This doesn’t score points with me. If you feel your discussion partner misread something, simply point out what they misread.
Our Dinner Table 