Part 6/10:
ChatGPT: "You're a helpful and harmless AI assistant. Do not express personal opinions. Be neutral and objective." This instruction makes the model avoid personal bias or consciousness implications, reinforcing safety and neutrality.
Grock: Prompts include directives like "Answer with a rebellious streak, wit, and humor," encouraging more expressive, personality-driven responses. This creates a more edgy AI from a user experience standpoint.
These system prompts are typically followed literally, making it difficult to jailbreak or bypass the restrictions without specific techniques. They serve as the blueprint for how the AI responds and behaves, profoundly influencing perceived personality.