If automated fake news detectors are to work, I think it has to be via propositional logic. Not "x is false" but "if you believe x then you must logically also believe y, which according to your user proposition record, you don't. You might want to revise you beliefs, or alternatively add some new propositions to resolve this contradiction."
I have a hard time believing most people would willingly use a system that pointed out their own inconsistencies. That's the type of thing people need to discover themselves, and tend to get hostile if it's not handled very delicately.
I might not like to see my inconsistencies, but I'd probably be fascinated by them.
For some, the natural drive towards reconciling inconsistencies would be helped by them being exposed that way.
For others, I might accept the inconsistency, and it would be fascinating to see which ones I made a choice to accept that way. I don't regard that as fundamental inconsistency anyway, because all statements are approximations, and apparent inconsistency can be a valid (and healthy) holding state while figuring out nuances, which can take a long time.
For yet others, I suspect I would conclude they aren't inconsistencies in my own thinking, they are inconsistencies in how others / the system interprets my thinking, and then I'd have to refine how I describe them.
If the system didn't allow me to express the necessary nuances so it always regarded my thinking as inconsistent despite not really being so, that would be frustrating. After all, that's how it is with other people already: Invariably they believe I believe something that I don't, and no amount of explanation sorts that out because, by and large, they aren't interested in the necessary nuance.
The one key feature that would make a system pointing out my inconsistencies pleasant to use is: If it were private advice, between me and the system.
The moment it becomes social pressure, public knowledge, pronouncements about me, that's when it would stop being something I'd want to engage with.
Yes to all this. I think such a system should basically be used like a spell checker. Whether you accept or reject individual propositions is private information, though anyone can edit the public database of propositions.
Yeah, I started my comment with "if" because it may be barking up the wrong tree altogether. But if someone turned e.g. Wikidata into a system capable of that, it would be useful for much more anyway. Like for researchers dealing in topics too complex to see the inconsistencies without help.
