Anyone remember the gpt-3 era where adding "this is important to my job" and "if this is wrong someone dies" to prompts made a measurable improvement in accurac

Anyone remember the gpt-3 era where adding "this is important to my job" and "if this is wrong someone dies" to prompts made a measurable improvement in accuracy and reduced hallucinations?