A methodology for verifying second-hand information — downgrade tweets / hot-takes / news / research notes / other AIs' conclusions to "hypotheses", decompose them into atomic claims, and verify each against primary sources with a confidence label. Use when the user says "verify / fact-check this",
A methodology for verifying second-hand information — downgrade tweets / hot-takes / news / research notes / other AIs' conclusions to "hypotheses", decompose them into atomic claims, and verify each against primary sources with a confidence label. Use when the user says "verify / fact-check this", "is this person right?", "what needs checking", "a tweet says", "another AI said", "is this rumor re