Generative AI as copywriter (part 1): pros and cons
Never worry about copywriting again?
Since the launch of ChatGPT, GenAI (generative artificial intelligence) has been a hot topic. No doubt, many of us have by now tried their hand at generating text and some of you may even have a GenAI model write texts on a regular basis. And why not? Their results are astoundingly good and sound very natural!
This might prompt many of us to wonder: with this technological advancement, do we even still need human copywriters at all?
Creative writing
First of all, some basics: where do we even find creative writing (or copywriting) these days? Here is a short (and definitely non-exhaustive) list of examples:
- Product descriptions
- Brochures
- Website content
- Blog posts
- Marketing content
- Social media posts
- And much more…
The human touch
Until recently, the creation of texts—especially the advertorial kind—was the sole domain of human copywriters. But with the advent of large language models (LLMs), which have learnt how to generate natural-sounding texts from analysing vast amounts of human writing and learning its patterns, this has changed profoundly. AI is, however, by no means a perfect replacement—as of yet, it is not capable of covering all dimensions of human intelligence. And that is why I would like to take the opportunity today to highlight some of the advantages and disadvantages that are inherent to AI and human copywriters respectively.
Advantages of human copywriters …
- Creativity – This is, as of yet, the domain of the human mind. GenAI creates texts based on data it has seen before, during its training. It therefore tends toward the average and not to unusual turns of phrase.
- Emotion – Emotional intelligence is another stronghold of humanity. Even experts currently disagree on whether this will ever change. This means, however, that only humans can effectively use emotional triggers in their writing—because they feel what their readership feels.
- Precise targeting of audiences – This is a much less clear-cut case. GenAI allows for a lot of adjustments through prompt engineering (using the “correct” command to achieve a certain output). But a human who has received a brief including a certain audience will be able to empathise better with this target group and will therefore likely produce a better result from the start.
- Culture, society and context – Humans are social creatures and each and every society has its own standards, expectations, values and ideas. A machine generating texts purely based on statistics will find this hard to understand. Humans, however, know their own culture and its intricacies inside and out.
… and their disadvantages
- Speed – Creative writing is a time-consuming process. Before the first few words can be put on paper, a copywriter will potentially spend hours on in-depth research. And once the text has been written down, it still needs to undergo polishing and feedback loops with the client before the final product can be released. GenAI, however, can produce good first drafts within seconds.
- Cost – It is hardly surprising that such a time-consuming endeavour is expensive. Using a generative AI model, which produces acceptable texts within seconds (and often free of charge or for only a few Euros per month), can be much more budget-friendly.
- Consistency – This usually only applies to larger or long-term projects which involve various people. Every copywriter has their own style, which can easily lead to inconsistencies. Using one GenAI model for all your texts might therefore lead to a more consistent style.
Generative AI as content creator
Some of GenAI’s great advantages are of course disadvantages of human copywriters that I already touched on above. I will not repeat those, in order to be able to shed some light on additional aspects.
Advantages of AI copywriting …
- Inspiration and ideation – This is less about asking GenAI to produce a text and using the unfiltered output. It is more about using it as a tool to support your writing process. GenAI can be very helpful in sparking ideas and defining the outcome or desired result of a text.
- No “empty page syndrome” – Many of us who have been tasked with writing know this problem: sitting in front of an empty page and not knowing where to start. GenAI can solve this problem quite elegantly: its output can be used as a basis to build on and avoiding staring at a blank page.
- Multilingual texts – Another very useful aspect of GenAI is its capability to generate texts in various languages. If you are looking to publish something in more than one language, you can immediately task your preferred GenAI tool with the translation (but beware, this has its pitfalls too; see my previous post about AI in translation). Or you might tell the AI to generate texts about the same topic in multiple languages. This does, however, pose the risk of the texts having significant differences in output as no GenAI model produces the same output for the same prompt twice.
- Data and research – One thing that machines definitely do better than humans is crunching large amounts of data in a very short amount of time. This could help you add some good data to your arguments. (But beware of hallucinations!)
… and drawbacks
- Hallucinations – After ChatGPT’s launch, the public quickly noticed that AI sometimes hallucinates, i.e. it invents things. Such as some legal evidence in the case of two unlucky New York lawyers who did not fact-check the AI output. So: beware of raw AI output! The advantage of human copywriters in this instance is that they base their texts on thorough research (or at least know when they make up things, whereas GenAI will insist on its output being correct, even if queried).
- Depth of information – GenAI needs a lot of data for its training. But this also means that it lacks special information and terminology for more niche topics. It is therefore not surprising that its text output often is rather superficial and provides little depth.
- Energy consumption – According to statistics, one AI prompt easily uses 10 times the amount of energy needed to do a standard online search via a search engine (see this Heise article). Even without considering the energy needed for its initial training! This raises the question of whether the use of generative AI tools is sustainable in the long term and how the global energy consumption will be impacted by the increased use of AI.
- Data privacy – If you have been following this blog for a while now, you will have noticed that this topic is very important to me. I have made this point before: a lot of providers (especially of free-of-charge tools) use the data inputs of users. What for? We do not know. So, if you ask GenAI to produce draft work emails for you, make sure to not enter any personal data or proprietary information.
Outlook
When discussing the question of who is better at copywriting, human or machine, it is not only their respective pros and cons we should be looking at. Just like in the case of human vs. machine in translation applications, we should reflect on who might be better in which situation. Meaning: when can I use AI to generate texts and when would it be better to ask a human copywriter for help? And what aspects should you be aware of when using GenAI? That will be the topic of my next post.
Comments are closed.