beaucronin

Knowing authorship has long been an important signal for judging work. Sometimes we specifically want authorship to be anonymous -- in peer review or contest judging, for example. But in many cases knowing authorship is useful. We are now in an era where that signal is (deliberately or not) obscured.

Here is a simple set of labels that I find useful for distinguishing how much AI contributed to a document or code file.

My pet peeve is that the IPCC reports are balled up in these massive PDFs. So I exploded the most recent one into an Airtable, with sections linked from the summary and so on. Still some cleanup, but it's mostly there. Interested if this is useful for anyone.