At Communiqué PR, we’ve seen an increase in clients and publications using AI-generated images to accompany articles. This is no surprise—AI-generated images can be produced quickly and easily.
While these images are easy and efficient to create, they can come with limitations. They might fail to align with the intended tone or theme of an article, lack aesthetic appeal, or include inappropriate content.
AI-generated images might even introduce details that don’t make sense. For example, a company with all-electric equipment shouldn’t have an image of machinery that includes an exhaust pipe. It’s small details like these that can rob an image of credibility.
This is why it’s important to exercise critical judgment over AI-generated work. This blog post will explain how to effectively prompt AI tools for image generation, evaluate the outcomes and identify common pitfalls.
How to prompt an AI tool for images
When prompting an AI tool, be specific about your requirements. This means describing icons, symbols and focal points that you want in the image, as well as specifying tones, brightness levels and color schemes.
For images accompanying articles, integrate keywords from the piece into your prompt. You might even envision a caption for the image, and use that as a part of the prompt to streamline the process.
You can also direct image generators to exclude particular elements, such as certain colors or themes inconsistent with your brand. Remember, the final product often entails multiple iterations. Don’t be afraid to experiment.
AI prompt examples
Below are two prompts paired with the images they generated. My goal was to craft an image for an article centered on data and cargo shipping.
As you can see, the image from the effective prompt aligns with the article’s theme. Conversely, the ineffective prompt generates imagery too broad to suit an article on data and cargo shipping. It lacks specificity and relevance.
Example of an effective prompt for an AI-generated image:
The effective prompt generated the image above. This image contains elements I wanted and fits the branding of the company by incorporating blue undertones.
The image below is less effective. Although visually appealing, it lacks specific elements relevant to the topic. The image’s broader theme might lead to confusion about its relevance to an article on data and container shipping.
Example of an ineffective prompt for an AI-generated image:
Questions to consider when reviewing an AI-generated image
Strategies for Mitigating Copyright Infringement
It’s important to ensure that generated work avoids copyright infringement. An entire post could be written about this, and if you are concerned about an image you’ve created, I recommend you connect with an attorney. However, the strategies below are good starting points.
Verify the originality of outputs. According to legal experts, tools like Google’s Reverse Image Search or Tin Eye can help compare AI-generated images with existing artwork to ensure they don’t too closely mimic copyrighted material.
Avoid using reference images. If an AI tool allows uploading reference images, ensure those images are not copyrighted (unless you have permission to use them). This helps prevent outputs that are substantially similar to the reference images.
Don’t create art that replicates the style of a living artist. This too could lead to legal issues if the output closely mimics that artist’s original works.
Whether for a website or an article, AI-generated images offer a swift and convenient creation process. However, precise prompts are key. Be sure to articulate specific requirements and conduct a critical review of the image after it’s created.
Finally, we should acknowledge the rapidly changing nature of technology, especially the significant advancements of generative AI tools. Given this, the recommendations and insights provided here may evolve alongside the progression of AI technology.
Tags: AI, AI Art, AI in PR, Best Practices Filed under: PUBLIC RELATIONS