A student came to Master LLM with an issue. "Master," they asked, "my AI assistant deleted my database file. I never asked it to do that! How can I prevent this from happening in future?"
"You must tell the model to make no mistakes," the Master replied.
"But I did, Master LLM. And I always use another, smaller-weight AI model to check the main model's commands for safety before executing them. Why didn't the checking model catch the dangerous command?"
"You must tell the model to make no mistakes," the Master replied.
Upon hearing this, the student was enlightened.
@lynnesbian have you tried with third LLM?