In OpenAI GPT4, you can give it a system prompt, like its personality. In Anthropic Claude, they instructed them to be, I think it’s… harmless, honest, and helpful. That order is important, I guess.
j previous speech k next speech