Which factor is most directly influenced by system messages during prompting?

• A. The model's architecture.
• B. The model's persona, tone, and safety constraints.
• C. The underlying training data.
• D. The size of the hidden layers.

Answer: (B)

System messages provide top-level directives that set how the assistant should behave in the current chat. They directly define the persona or role, the tone, and the safety and content constraints that guide every reply. Because these are instructions delivered at prompt time, they shape output style, politeness, formality, and what content is allowed or disallowed in that session. The model’s architecture, its training data, and the size of its hidden layers are fixed properties of the model itself and do not change with prompting. Prompts can steer how the model behaves within those fixed constraints, but they don’t alter the underlying structure or knowledge. So, the factor most directly influenced by system messages is the model’s persona, tone, and safety constraints.