What she is saying is correct about the utility of LLMs in scientific research though.

When your user says that your product doesn’t work for them, saying they’re using it wrong is not an excuse.