With the rapid rise of generative Artificial Intelligence (AI), the writing process will never be the same again. For those in the creative space, this can feel both exciting and scary. Some are intrigued by the prospect of faster and easier content production. However, others worry that these tools may compromise the creative process and result in mediocre work, potentially taking away jobs from the copywriters, content strategists, and marketers valued for their sophisticated writing skills.
At Aquent Studios, we believe AI can support and augment creatives and the skilled work they deliver. But the only way to fully understand how AI would work in real-world scenarios is to put it to the test. That's why our team of Copywriters and Content Producers spent months testing the strengths and weaknesses of various AI tools for writing marketing copy.
The result? We found that AI excels at drafting long-form content, repurposing written work into different formats, and improving quality—led by an experienced writer, formulating the prompts and shaping the outputs. But when it comes to short-form marketing copy like social ads, web copy, or taglines, current AI technology frequently falls short.
Read on to discover how we came to these conclusions and what these results mean for content and creative teams.
How does generative AI write content?
To find out how AI writing tools work, let's head straight to the source. ChatGPT gives us the following explanation:
“Generative AI writing tools like ChatGPT and others utilize advanced algorithms and deep learning models to generate human-like text based on the input provided. These tools are trained on vast amounts of data, including books, articles, and websites, to learn patterns and language structures. During training, the models predict the next word in a sentence based on the preceding context, enabling them to understand and generate coherent text. When a user interacts with a generative AI writing tool, such as ChatGPT, they provide a prompt or ask a question. The model then processes the input, analyzes the context, and generates a response based on the patterns it has learned from the training data. The output is an approximation of what a human might say, although it may not always be perfect or accurate. Continuous feedback and iteration are crucial for refining these models and improving their capabilities over time.”
Knowing how generative AI produces its outputs is revealing. These technologies use an input (also known as a prompt) in order to figure out what to write and then predict the next most likely word in the sentence based on the words that precede it.
A few points here are critical. What you provide as a prompt drives the outputs from the model, in sometimes unpredictable ways. From our experience, it's not just one-and-done. Crafting your prompts to get the content you're looking for takes iteration and refinement. You get better at it as you practice, but you're also not usually satisfied with the first output you get. You might want the output to be shorter or longer, more or less technical sounding, more conversational or more formal, and on and on. You might need to try a few different prompts or write the content yourself and have the AI rewrite it for you. What it all comes down to is that these AI tools still require a human to drive them—providing the direction and coaching to get the best out of them.
Another point to highlight is that generative AI uses probabilities to determine the next word that makes the most sense, based on its understanding of human language and all the data used to train its model. That's why these tools appear to be typing as word by word appears across the screen. While these algorithms are what make the text sound natural and human, they can have both advantages and disadvantages when it comes to different types of marketing content.
Benefits of generative AI for writing long-form content
Drafting rough content
There's no doubt that AI-powered writing tools can streamline the writing process. For example, AI performs well if you use it to kickstart an initial writing deliverable. You can give it an outline or key points, and it can help you form a rough draft, or it can help you write or rewrite a particularly tricky area. Then, you can focus on refining the outputs into a final product. AI can also generate various options that give you the perfect jumping-off point, which helps whether you're facing writer's block, are under a time crunch, or just need a push to get the ideas flowing.
In one recent experiment, we wanted to produce a 45-second script for a teaser video voiceover. Using a video script template in Jasper.ai, we described the video, adding background information that the script should cover and adjectives that describe our tone of voice. The results had several solid outputs, including an impressive draft that we felt was 80% there. With some light editing that pulled in the best elements from other output options, tweaked the language, and adjusted the flow, we produced a final script that we could give to our voice-over talent in an hour from start to finish.
Editing content
AI can also assist in the revision process, providing suggestions that range from simple proofreading to restructuring sentences or paragraphs to improve flow and readability. With their ability to analyze text for grammar, syntax, and clarity, AI writing assistants can quickly suggest changes that will enhance the overall quality of a piece of writing. Using the business or enterprise account features of tools like Grammarly or Quillbot takes it one step further, allowing users to customize those recommendations for their brand to match their company's style guide and tone of voice.
Repurposing content
Another area where AI can be helpful is repurposing content into different formats or channels. AI's ability to rework and condense existing text is impressive. It summarizes copy well and can consistently get to the essence of a piece of writing. AI can take a longer deliverable and rework it into new versions for different channels. For example, an AI tool can condense a detailed campaign story, press release, or blog post effectively. A writer would then take this condensed version for a landing page intro or social media copy, adding the necessary channel nuances.
Overall, AI-powered writing tools provide actionable feedback that helps skilled writers create polished long-form content faster and easier than before. These tools allow writers to focus on developing and shaping their ideas and refining their work rather than spending time on writing mechanics. They also ensure the deliverable meets essential requirements, like aligning to brand style guides, hitting a certain character count, and including specific keywords for search engine optimization. Generative AI should not be relied upon to replace human writers—their business judgment, institutional knowledge, unique perspectives, and humanity. Instead, they should be considered valuable new tools in a writer's toolkit.
Challenges of generative AI for writing short-form copy
Unlike our experiments with longer-form content, we discovered that the current iteration of AI struggles with short-form copy.
When conveying an idea in a few snappy words, originality is essential, and every word counts. Great copy is inventive, powerful, and attention-grabbing while at the same time being simple and clear. It takes years for copywriters to hone this skill and learn to write like this, and it's where AI faces challenges. Since AI uses probability to predict the next word, it is designed to produce the most obvious or typical phrasing to complete a given sentence or thought, often bordering on generic or over-used language to achieve this goal.
Take, for example, a recent line of copy promoting Prime on an Amazon package. The line goes, “Whether you're a dog person or a Doja Cat person, it's on Prime.” This play on words from “whether you're a dog person or a cat person” took the trending topic of Doja Cat dressing as a cat at the Met Gala and used it to make a clever and current idiom. AI is not capable of “thinking” this way, at least not yet.
For comparison, when asked to create a few short sayings to advertise the variety of options available on Amazon Prime, this is what AI came up with:
- “Endless possibilities, at your doorstep.”
- “Unleash the power of choice.”
- “One-stop shop for all your needs.”
- “Variety is our specialty.”
- “Explore, discover, and get it all.”
- “Experience the ultimate shopping convenience.”
- “Discover a world of options.”
- “Find everything you need, all in one place.”
- “The ultimate shopping destination.”
- “Limitless options, limitless potential.”
As we previously mentioned, working with AI is more complicated than creating a prompt and publishing the output. After initial prompting or inputting a base to be reworked, you need to guide the AI through feedback rounds and edit the output to say precisely what you want to say. This process can be effective and save you time when dealing with 500 or more words, where the AI could help restructure and correct a lot of content, but not so much for 150 characters or less.
But before you throw out that list of 10 generic ideas above, consider this. As copywriters, we at times intentionally brainstorm the “bad/obvious/throwaway” ideas to get them out of our system, which allows us to push beyond into other territories. Even coming up with these bad ideas takes some time. So writers are using AI to create these brainstorms for them as a “don't do this” list. That means we can much more quickly jump into creative exploration, avoiding AI's cliche ideas.
Another area where AI is limited is in its ability to adapt its tone of voice in short-form copy. With less to work with, every word is a careful choice. Even when summarizing a pre-written base for another channel, AI tools lose some tonal nuance and subtlety in the initial output. That means text typically needs rewrites to suit a particular tone of voice, target audience, and channel requirement.
Generative AI is a powerful writing tool for some types of marketing content
At Aquent Studios, we are excited about the continued development of AI. These tools have come a long way in recent years and are now capable of producing high-quality content that is engaging and informative. However, despite their impressive abilities, we believe that writing AIs will never replace the need for subject matter experts or copywriters.
While AI can produce well-written content, it lacks the human touch that comes from an expert's knowledge and experience. Subject matter experts possess unique insights in their field that an AI cannot replicate. Copywriters, on the other hand, bring creativity and personality to their work that AI-generated content cannot match.
While certain aspects of AI may change the way we write, original thought, a distinct voice, and emotional appeal will always be hugely important. Generative AI can certainly be useful, but it should be viewed as a tool for expert human writers rather than a replacement for them.
This blog talks about generative AI tools and their potential applications. As these technologies rapidly evolve, we encourage you to research the latest developments in terms of their capabilities, safety and security, and ethical use. For more information on the responsible use of generative AI, download our whitepaper, “Using Generative AI for Design: Legal Considerations and Best Practices.
