Its mistakes aren’t random. They’re the machine showing through.
You can now work out what an AI will get wrong before it gets it wrong. Here are the
three classics. Each one is caused by something you have already seen, including the model’s
knowledge cutoff.
It is slow to the news, and it does not know that it is. Everything after the day its reading stopped is simply not in the shell.
1Pick a question it fails
Do thisRead its answer. Then press Why does it do that? The reason is
always something from an earlier lesson.
