The power of ChatGPT prompts: best practices for effective content generation

Working at Jasper AI has provided me with invaluable insights into the process of creating effective prompts to achieve optimal AI-generated text outputs, particularly in the realm of marketing. However, the best practices that I have learned so far are not limited to marketing alone, as they can be applied to a wide range of topics.

It’s still your job

It is important to remember not to relinquish complete control to tools like ChatGPT as if you are no longer the one steering the ship. This may be tempting, especially in the early stages of working with AI-generated text, as it can be difficult to discern where your role ends and the AI's begins. However, it is crucial to maintain a reliable workflow that guides the AI to assist you, rather than the other way around. In other words, you should be the one directing and supervising the AI-generated text, not the other way around.

In this post, I’ll be sharing my insights gained through recent immersion in the AI text-to-text and text-to-image generation space. However, it is important to note that breakthroughs and new releases in this field are happening at a rapid pace, so my learnings will likely only scratch the surface of what is to come.

A typical blog post workflow

Here is a helpful list of the steps that many writers follow when creating a blog post and it also happens to be a popular use case for marketing professionals. Utilizing ChatGPT doesn't necessarily change your workflow, but rather accelerates the process and can sometimes enhance your intuition.

1. Brainstorming ideas: Before starting to write, you need to come up with a topic that is both interesting and relevant to your target audience.

2. Researching: Once you have a topic in mind, you need to research the subject matter thoroughly.

   • Note that using AI generation tools for research can be a rapid way to begin moving in the right direction. However, it is important to keep in mind that human analysis and evidence must be incorporated to ensure that the output is relevant, factual, and accurate.

3. Outlining: After conducting your research, create an outline for your blog post.

4. Writing the post: With your outline in place, begin writing the first draft of your blog post. Start with an attention-grabbing introduction, followed by the main body of the post and a conclusion that wraps up your main points.

5. Editing and proofreading: Edit your content for clarity and concision, ensuring that your post is easy to read and understand.

   • Note that Grammarly is a useful app for editing and proofreading things like grammar, spelling, and punctuation errors.

6. Formatting: Once you have polished your content, format your blog post with headings, subheadings, and bullet points to make it visually appealing and easy to skim.

   • This is still best done using your regular text styling tools in your productivity app of choice.

7. Adding visuals: To enhance the visual appeal of your post, include relevant images, videos, or infographics that help to illustrate your points.

8. Publishing: Finally, publish your blog post on your website or blog. Share your post on social media and other channels to help promote your content and reach a wider audience.

   • While the actual publishing process is not yet AI-assisted to my knowledge, using AI to repurpose your blog post for social media content works really well and is often the next step.

Commands + context are key

For me, the breakthrough was learning how to use commands and provide context to tell the AI exactly what I wanted. Similar to humans, AI requires context to generate optimal results. It is important to consider precisely what you want the AI to produce and provide a prompt that is tailored to that specific objective.

You can think of the prompt command as a combination of an action verb + the content type + a descriptor.

Take this basic prompt as an example:

“Write about navigating ambiguity."

Here is a better prompt using the full formula ie the action verb, the content type, and a detailed descriptor:

"Write a blog post about the importance of navigating ambiguity for complex projects."

By providing clear and specific instructions, you can ensure that the AI generates relevant and accurate output, which can then be used as a starting point for further editing and refinement. Therefore, when working with AI, it is crucial to provide context and direction to ensure that it produces the desired outcomes. At this point, it should be super validating that you still need to be in charge and actively tweaking the context by observing the output.

Brainstorming ideas

AI generation can be an excellent tool to help humans overcome writer's block when starting from a clean slate. When a writer is stuck and has no idea how to begin writing, AI-generated tools can readily generate suggestions or prompts (not to be confused with your input prompt) to kickstart the writing process. These prompts can include keywords, ideas, or even whole sentences that can inspire the writer and lead to new ideas.

Examples of commands for brainstorming ideas:

1. “What are some creative ideas for [insert topic]?"

2. "Can you suggest some unique approaches to [insert problem/issue]?"

3. "What are some insights or perspectives on [insert topic/issue] that people may not have considered?"

Remember, the more specific and targeted your prompts are, the better results you will get from the AI-generated output.

Researching your topic

Here are some examples of commands you could use for researching a topic.

1. "What are the key features/benefits of [insert product/service]?"

2. "What are some current trends or emerging technologies in [insert industry/field]?"

3. "Can you provide some historical context for [insert event/topic]?"

4. "What are some different perspectives on [insert issue/topic]?"

Remember that while AI-generated text can provide a starting point for research, it should be used as a supplement to, rather than a replacement for, human analysis and research.

Outlining your article

Here’s the command I used for generating a blog article outline:

“Create a blog post outline on the importance of navigating ambiguity for complex projects”

This was the output from ChatGPT:

I. Introduction
A. Definition of ambiguity and its relevance in complex projects
B. Importance of navigating ambiguity in complex projects
C. Overview of the key points covered in the post

II. Understanding ambiguity
A. Definition of ambiguity in project management
B. Examples of ambiguous situations in complex projects
C. Potential risks and consequences of unresolved ambiguity

III. Navigating ambiguity
A. Strategies for identifying and clarifying ambiguous situations
B. Importance of effective communication and collaboration in navigating ambiguity
C. Tools and techniques for managing ambiguity in complex projects

IV. Benefits of navigating ambiguity
A. Improved decision-making and problem-solving capabilities
B. Enhanced project outcomes and reduced risks
C. Increased adaptability and agility in a rapidly changing environment

V. Challenges of navigating ambiguity
A. Common obstacles to navigating ambiguity in complex projects
B. Ways to overcome challenges and maintain progress
C. Potential drawbacks of overreliance on navigating ambiguity

VI. Conclusion
A. Summary of the key points covered in the post
B. Final thoughts on the importance of navigating ambiguity for complex projects
C. Call to action for readers to implement strategies and techniques for navigating ambiguity in their own projects.

Writing the blog post

Next, I provided the following command + context:

“write me an introductory paragraph for the above outline”

This was the output:

Navigating ambiguity is an essential component of managing complex projects. In any project, there are often unforeseen challenges, unexpected roadblocks, and unclear paths forward. When faced with ambiguity, project managers and teams must be equipped with the skills and strategies necessary to identify and clarify ambiguous situations, and effectively communicate and collaborate to reach a resolution. In this blog post, we will explore the importance of navigating ambiguity for complex projects, including strategies for identifying and managing ambiguity, benefits of effective navigation, and potential challenges and drawbacks. By understanding and embracing ambiguity, project teams can improve decision-making, enhance outcomes, and increase adaptability in an ever-changing environment.

You can then step through each of the items in your blog post outline by using something like this:

“Write a paragraph on the above IV. Benefits of navigating ambiguity”

…which produced this:

Navigating ambiguity can lead to numerous benefits for complex projects. By effectively identifying and managing ambiguous situations, project teams can improve their decision-making and problem-solving capabilities. This includes the ability to quickly adapt to changing circumstances, identify potential risks and opportunities, and respond in a timely and effective manner. Effective navigation of ambiguity can also enhance project outcomes and reduce risks, by ensuring that project goals are clearly defined, and that all team members have a shared understanding of project objectives. Additionally, navigating ambiguity can increase adaptability and agility in a rapidly changing environment, enabling teams to pivot quickly and adjust course as needed. Ultimately, by embracing ambiguity, project teams can improve their overall performance and increase the likelihood of project success.

So this paragraph did not resonate with me as it feels a bit shallow and vapid like a bunch of keywords were thrown together and regurgitated back at me. In this case, it’s essential to take control of the situation and use the AI-generated paragraph as a good starting point, while also contributing your own ideas and insights.

My next steps would be to explore the use of different commands and contexts, to generate the other sections of the blog post to tailor them to my satisfaction. Hopefully, this provides some encouragement that AI generation is augmenting your skills and expertise, rather than replacing them.

Adding AI generated visuals

AI image generation is still a bit tricky, but lots of fun if you can manage the current limitations.

I think it’s safe to say that AI generation tools are not yet sophisticated enough to comply with a brand identity consistently and predictably. I’m sure that this day is coming though. In the meantime, we can exploit the limitations of text-to-image tools to our advantage. For example, AI gen is pretty reliable at generating background images over which you can layer your brand assets. And of course, the more spontaneous and unrepeatable, the better so if your campaign involves original abstract art you’re in luck as far as leveraging text-to-image AI tools.

Mostly I have been using Midjourney for text-to-image AI generation. Dall-E 2 is another brilliant text-to-image tool that introduced the concept of outpainting. While the following prompt tips happen to be for Midjourney, the general concepts would probably apply regardless of the tool being used.

Once again we explore the anatomy of the prompt, but this time to describe the visual to be generated. The anatomy of a visual prompt can start with the subject followed by the attributes of the subject. You can add a style broken down into two sections: first the artistic attributes, then the render attributes. Tacked on to this are the Midjourney specific parameters such as --aspect, --no, --creative to name a few. Your prompt doesn’t have to be complicated, the idea is that you have all of these prompt elements at your disposal to help you get as close as possible to the image you want.

You can also browse the Midjourney community page which showcases many images along with the prompts that generated them. You can then copy any of those prompts directly to experiment with. You can also use a tool such as MidJourney Prompt Helper that will help guide you through building out your text-to-image prompt.

The push-pull nature of this collaboration

I’m definitely aware of the push-pull nature of my collaboration with AI generation tools, especially text-to-text generation. There’s usually a bit of inertia that sets in when tools like ChatGPT have been doing a lot of the heavy lifting. It’s almost annoyingly frustrating to take back the reigns I think because you have to reassemble the thing in your own mind again, especially if it’s a complex workflow. The act of reinserting yourself back into your own work can sometimes feels like a chore. I try to dismiss this as laziness with a good reminder that I’m responsible for the quality of this content.

AI can be a key player on your dream team

The nature of this relationship has rules and guidelines that must be adjusted to and perfected over time just like any other relationship in your life; you get what you put into it. By learning a little about what the AI needs such as crafting more thoughtful prompts and using context to iterate when needed, you can refine the output until it meets your expectations and delivers the level of quality that you desire for your projects.

Don't be afraid to take control and use AI generation tools, while also contributing your own ideas and insights. With the right combination of human expertise and AI technology, you can evolve this relationship into a rewarding and powerful collaboration for just about any genre of content or idea generation required for your marketing team and beyond.