What is the future of tags supposed to be? (in a world where you might want “AIs automatically to tag things”, and/or “AIs solve the problem that tags were previously mostly solving some other new way.”)
A simple world is “when posts are published, or, periodically when they are saved after an author has written like 500 words, some AI process searches over the existing tags and makes some tag suggestions, which get the Robot icon until someone actually upvotes them.
But, it’s possible it’s supposed to change more radically like that.
Like, maybe AIs suggest new tags. Maybe we attempt to have an .MD file that’s like the comprehensive guide to our tagging philosophy and also LessWrong’s ontology, and maybe that file is wiki-editable.
Maybe we are supposed to have it build “Grokipedia but good.” Or, maybe AIs can’t be trusted with the subtleties of summarizing, but, maybe they can automatically assemble extracted quotes from prominent posts from a given tag.
I find myself particularly curious about @abramdemski’s thoughts on “have an AI prompt file that’s a comprehensive guide to tagging philosophy and the LW ontology”, I think you had previously written some stuff about those types of things back when Tagging first came out.
What is the future of tags supposed to be? (in a world where you might want “AIs automatically to tag things”, and/or “AIs solve the problem that tags were previously mostly solving some other new way.”)
A simple world is “when posts are published, or, periodically when they are saved after an author has written like 500 words, some AI process searches over the existing tags and makes some tag suggestions, which get the Robot icon until someone actually upvotes them.
But, it’s possible it’s supposed to change more radically like that.
Like, maybe AIs suggest new tags. Maybe we attempt to have an .MD file that’s like the comprehensive guide to our tagging philosophy and also LessWrong’s ontology, and maybe that file is wiki-editable.
Maybe we are supposed to have it build “Grokipedia but good.” Or, maybe AIs can’t be trusted with the subtleties of summarizing, but, maybe they can automatically assemble extracted quotes from prominent posts from a given tag.
I find myself particularly curious about @abramdemski’s thoughts on “have an AI prompt file that’s a comprehensive guide to tagging philosophy and the LW ontology”, I think you had previously written some stuff about those types of things back when Tagging first came out.