A company finds a foundation model producing images that do not match the prompts. They want to adjust prompting techniques to reduce irrelevant images. Which prompting approach directly helps exclude undesired content?

Negative prompts explicitly tell the model what not to include or generate in the output. By specifying undesired content, the model can actively avoid producing irrelevant images, directly addressing the company's problem. Zero-shot prompts are a general approach where the model performs a task without specific examples, which doesn't inherently exclude undesired content. Positive prompts describe what should be in the image, but don't prevent the generation of unwanted elements. Ambiguous prompts would likely worsen the problem by providing unclear instructions, leading to more varied and potentially irrelevant outputs.