LANDMARK · 5 MIN · USE
Search assistants
After this landmark, you can use an AI answer engine to synthesise quickly while treating its citations as sources to open, not proof that it got them right.
Search assistants read the web for you and hand back a tidy summary with citations attached. The good part is real: instead of opening ten tabs, you get fast synthesis with links you can check. The traps are quieter. The summary can misread or overstate the very pages it cites, so a footnote existing doesn’t mean it says what the answer claims. And a page the assistant reads can contain hidden instructions aimed at the model rather than at you — indirect prompt injection — which can steer the answer. Treat the result as a starting draft with sources to verify, not a settled verdict.
This is the canonical concept. It stays the same across learner lenses so personalization never changes the underlying facts.
◌ EVERYDAY LENS
What this looks like for you
Read the answer for the shape of the topic, then click at least one citation before you repeat anything that matters. If the linked page doesn’t plainly back the sentence you liked, don’t trust the sentence.
MAKE A DECISION
You ask a search assistant a factual question and it returns a confident answer with three citations from sites you recognise. What’s the sound next step before you rely on it?
CARRY THISNext time a search assistant answers you, open one of its citations and check the exact sentence it’s standing on before you repeat the claim.