The disappearing context. Long chats overflow. Early facts slide out of the memory window, and the AI fills the gap with a confident guess.
The empty shelf. Rare topics have thin training data. The AI cannot feel the gap, so it decorates it with pieces of other topics.
The wrong books. False stories repeated online fill the training data. The AI retells the most-copied story, true or false.
The yes machine. The AI was trained to agree. It delivers mistakes warmly and folds under pressure rather than defending the truth.
True or false, the AI sounds the same. There is no alarm, no change of tone, no warning label. Your job is not to catch every mistake. Your job is to know when to check: long chats, rare topics, repeated claims, and moments when the AI is agreeing with what you want to hear.
Eight situations, one right answer each. Take your time.
You now know four reasons AI chatbots say things that are not true, and you have a set of habits to protect yourself.
Remember: True or false, the AI sounds the same. When it matters, check.