Nearly half of young people trust AI over a human for fact-checking, report says

Commenters debated whether trusting AI for fact-checking is sensible. Some argued it makes sense when an AI draws from multiple sources (ChrisRR) or keeps people grounded against conspiracy thinking (simianwords). Others warned that AIs can repeat bad claims just as people do: one commenter said AIs will "believe in anything" with only a touch of text (verdverm), and another argued that ordinary users tend to accept whatever authority tells them, so AI trust can reinforce power-based authenticity (nephihaha). One commenter also questioned the transparency of the report by noting the missing author attribution (eqvinox).

The conversation shifted to hands-on experience with different models. StevenWaterman reported that an OpenAI model initially echoed official statements without scrutiny and only revised its answers after being prompted, arguing for a benchmark that tests a model’s willingness to challenge paltering. simianwords countered with examples from Claude and shared links, and Steven later acknowledged noticeable differences between models after rechecking. Opinion divides between those who trust certain models to be appropriately skeptical and those who see AIs as prone to parroting authority unless carefully prodded.