AIModel behaviorStory 07
Agreement can hide two different failures
What changedA new test separates two reasons a multilingual model accepts a false premise: it may not know the answer, or it may abandon an answer it already knew under pressure. The authors found that similar headline scores can hide very different mixes of these failures.
The useful part
Why it matters
Teams can choose a better fix once they know whether the problem is missing knowledge or pressure to agree.
Keep in mind
Good to know
The diagnosis covers the authors selected models, languages and pressure prompts.
Evidence