Generative AI as copywriter (part 2): use cases and practical advice
GenAI as copywriter: advantages and disadvantages
Last month, I looked into the pros and cons of both generative artificial intelligence (GenAI) and human copywriters and shed some light on aspects such as creativity, cost and consistency.
Unfortunately, this left little space for another important question: which option is better suited in which setting. When would it be recommendable to work with a human and how do I use GenAI for my copywriting needs?
Use cases
Decision aid: artificial or human intelligence?
In the course of my research for this short blog post series, I came across a blog about GenAI in copywriting. And from it, I would like to share this interesting quote with you:
“Think of an AI copywriter as an extremely intelligent, 12 year old child that knows a vast amount about the world from the internet.” (Found in https://www.hypotenuse.ai/blog/ai-copywriting-explained).
In my opinion, this quote illustrates quite well what AI can currently do and what it can’t do (yet?). Due to its vast amounts of statistical language data and huge processing power, it is very good at formulating sentences and creating texts about a variety of topics. Just like a 12-year-old could. And because it was trained with data from the internet, it has access to a whole host of information about our world, which means its output does contain some useful information.
What it is not very good at, however, is creating texts that go beyond the obvious or that look into niche topics, that argue for novel points of view, make connections that have not been made before or use very original language. These are capabilities which are, as of yet, unique to the human mind.
Considering the above, I personally think that the following division of tasks between human and machine would make sense:
AI use cases
I think it is safe to say that no human copywriter will ever beat GenAI in speed and cost. If your goal is to create large amounts of content which is on the superficial side of things, GenAI is your friend. A few examples for great AI use cases:
- Social media posts
- Blog posts containing general or superficial information
- AI for sparking ideas: ask the GenAI tool of your choice to generate a text about a certain topic to get ideas for content to include. I actually did this for this short blog series and ended up adding one or two arguments which I probably would have overlooked otherwise!
The most important rule when using AI generated content is, however, to never just copy and paste the output. Always make sure to check and edit thoroughly; but we will talk about this in more detail later in this post.
Human copywriters
GenAI is good at generating high-quality text—especially if you know how to use it (see the practical advice section below).
But even so, humans are still a vital part of the content creation processes. A lot of this is down to the fact that whenever you would like to address the feelings and needs of a certain audience, humans can pour their empathy into these texts to make them as appealing as possible. Examples include:
- Posts or articles which require in-depth research into complex or niche topics.
- Especially in cases where these texts contain explanations adapted to the intended audience and their level of knowledge about the topic.
- Texts addressing readers on an emotional level, painting a positive picture of brands or intended to convince readers to buy a product (or other calls to action).
Another issue with GenAI is data privacy. Should you really hand company-sensitive data to one of the modern-day tech giants? When working with a human copywriter, you at least have the possibility to have them sign a non-disclosure agreement as legal remedy in case of a leak—but do we know what the companies whose tools we feed with our data do with it? No, because they tend to be very reluctant to share information on this…
Cutting it down to the essentials
If you are still unsure about whether GenAI would be an option for you or whether to commission a human copywriter, ask yourself the following question: what impact will this text have on my brand?
The greater the impact, the better it is to work with a human.
Especially in cases where you would like a text written in a language you don’t speak, a human copywriter can tell you exactly what it says. If you use GenAI, you can of course tell it to translate the text, but you will never know whether the translation is accurate and reflects all the nuances of the originally generated output (you can find more about this issue in this post).
Practical advice for using GenAI
So, you have decided to use GenAI to create some of your content. What are things you need to look out for?
Tasking
The quality of AI output is hugely influenced by the quality of the commands (prompts) entered. They tell the tool of your choice what it is you want it to do. The better your prompt, the better the AI can calculate the most appropriate result.
Biases
Many people think machines are neutral. Unfortunately, this is not the case. GenAI is based on large amounts of language data taken from the web, which leads to the following problem: from the human-made content on the web, it learns not just language patterns, but all our biases and prejudices, which are then reflected in its output. And most of us are probably aware of the fact that our modern society still has a long way to go regarding equality—including the way we speak and write.
Processing power
Like most software programs these days, GenAI can be used in two ways: via offers on the web or buying a locally operating model. Such local models offer the advantage of data privacy and potential further training and customisation. They are, however, extremely hungry for processing power. To be able to run your own local AI, you need very powerful machines, which also consume a lot of electricity.
Workflow
You would now like to get started and generate high-quality texts using GenAI. To be able to achieve a good result, I recommend following this 9-step workflow:
- Think about the topic you would like to write about (just like you would if you were to write the text yourself or ask a copywriter to do it for you) and what you would like to achieve with this text. Take note for your prompt.
- Choose an appropriate GenAI tool. Not all tools are created equal, some focus on certain languages or text types.
- Open the prompt line and start entering your prompt. Make sure to assign a role to the GenAI tool of your choice, e.g. “You are a professional copywriter.”
- Provide context—e.g. “A space company asked you to create a blog post for them.”
- Provide instructions about topic and length of the text. In this section, you can include examples for points you would like the text to discuss.
- If you like, you can also add some additional requirements; e.g. telling the tool to use a certain author’s style or certain features (such as plain language).
- Optionally, you can also add formatting instructions.
- Generate output.
- Check, revise and edit the output until it fulfils your expectations. Or adjust your prompt and repeat the process to refine the output.
Some final thoughts
Over the last few months, while looking into AI in translation and text creation, there is one thing that has become increasingly clear to me (and hopefully you as well): while AI is undoubtedly an incredibly powerful tool that simplifies many tasks, it is not always the best choice.*
So, what do you need for a successful text? No matter if it is created by a human or GenAI: you need to have a good idea of what you would like to achieve with it, both for briefing a human and prompting AI. To be able to use AI effectively, you should also look into prompt engineering. i.e. how to phrase prompts to achieve optimal output. But either way, you will need time for feedback loops, both for AI and human copywriters.
Do you still have questions about whether or not to use GenAI or machine translation in your language workflows? If so, please feel free to contact me. I would be more than happy to advise you on the optimal use of both human and machine.
* Another aspect to consider here: AI and machine translation (MT) are most powerful in what we call the “big languages” (languages with a large number of users). The reason for this is that AI and MT training needs a lot of high-quality data to be able produce good output; this amount of data is often not present in languages with a relatively small number of speakers. Therefore, modern technology tends to put them at a disadvantage.
Additionally, not all AI tools are available in those big languages (such as Spanish, French, German or Chinese). Some models are only available in English, which means that users have to have a very good command of the language to be able to prompt effectively. And in some cases, access to non-English versions is only available for a fee.
Comments are closed.