A company uses Amazon Nova Canvas to generate images and needs to prevent certain items from appearing in the outputs. Which technique should they use?

The correct technique is to include a negative prompt specifying items to exclude. In generative AI models like those used in Amazon Nova Canvas, a negative prompt explicitly tells the model what not to include in the generated output. This is the most direct and effective way to prevent unwanted elements from appearing. Increasing the temperature value would make the output more random and creative, but wouldn't specifically exclude items. Writing a more detailed positive prompt might implicitly reduce the likelihood of unwanted items, but it's less precise than a negative prompt and doesn't guarantee exclusion. Switching to a different foundation model is a drastic and often unnecessary step; most models support negative prompting for this purpose.