This is a pretty interesting take. If AIs have as their main knowledge base Internet contents, they're basically roleplaying as a human doing research online - they also need to be thought to think critically about whether some information _feels_ reliable or not. Are current LLMs capable of _really_ doing critical analysis or do they just pretend being able to do it?