Training AI on reliable data is integral. We don’t want to see AI doing the same things it does in ChatGPT where it will make up an answer if it isn’t sure of the answer.