When ChatGPT took the world by storm in 2022, people were quick to predict that it would render copy & content writers obsolete. Why would a company pay hundreds of dollars for a blog post if it can just prompt an artificial intelligence (AI) tool to spew out the article in seconds?
Today, anyone who has tried to go down this route knows why – current AI tools suck at creating GOOD content. Sure, the quantity is there, but quality is almost always lacking.
AI-generated content is typically:
- Bland – AI works by chaining the statistically most relevant words in response to a query. What you get is a soulless average. It's void of emotion and brand voice both in terms of content and style.
- Prone to errors – It's common for AI to make up information or confuse different contexts, producing misleading and sometimes blatantly incorrect content. Worse yet, these mistakes might be difficult to identify unless you're an expert in the field.
- Recognizable – Phrases like “In today's dynamic landscape” and stylistic choices like when the conclusion brutally reiterates previous points are telltale signs that the content is AI-generated, which tends not to sit right with readers.
- Difficult to edit – AI struggles to understand the broader context of your request. So when you ask it to edit some part of the text it has created, it often changes style, duplicates existing content, messes up the flow of the piece, leaving the editor to have to do it all manually anyway.
- A plagiarism risk – since AI tools often regurgitate existing content, there's a high chance that the AI's creation already exists somewhere else, potentially landing you in hot water with the original creators.
That said, prompt engineering, which is the skill of asking AIs the right query to get the best response, can minimize – but not eliminate – some of the issues. By specifying the style and the context, by providing previous examples, by pre-empting potential issues it's possible to dramatically improve the quality of the content.
Even then, experience shows that the content remains subpar. Plus, writing high-quality prompts is such a time-consuming process (especially since you have to write multiple, as the AI is likely to get it wrong at first) that the time-saving benefit of using AI begins to fade.
And yet, my colleagues and I find ourselves using AI tools every day – specifically for content creation.
Here are seven ways how we use AI at Truesix.
1. Research
As a company working with tech clients and startups, the range of topics we write about is very broad. One day we'll be putting together a piece about micro mobility legislation in Paris and then deep dive into silicon wafer supply chains and production the next.
AI tools have proven to be extremely useful for researching a topic because you're able to quickly ask follow-up questions, as well as have it explain things to you in simpler terms. Before, one of the biggest issues was formulating the query for Google and then having to sift through endless articles to piece together a complete picture.
With ChatGPT, the foundational research gets accelerated dramatically. That said, any specifics must be cross referenced and anything you end up putting to paper must always be double checked. Whereas tools like Perplexity allow you to supercharge your Googling.
2. Translation
Being able to specify translation context and style makes AI tools some of the best translators out there (as long as it's to English). Moreover, you can ask questions about the translation and request AI to optimize the text. All this pre- and post-translation manipulation allows you to get more out of the translation, making these tools particularly valuable when translation is done for research purposes.
Moreover, tools like ChatGPT help tackle another MASSIVE problem. Bilingual writers are forever cursed to think of the perfect term/phrase/idiom in one language only for it to be nigh-impossible to transpose to another. But with ChatGPT's complete library of English idioms and phrases, finding a suitable translation is easier than ever.
3. Proofreading
At Truesix, we always review each other's work before sending it off to a client. However, we're also perfectionists, so we usually try to get the first draft as polished as possible (and because it minimizes the editing effort colleagues have to put in). That's where AI can come in handy – for a first round of proofreading.
Running the article by an AI tool, we can identify areas for improvement, weed out simple errors, and evaluate the clarity of argumentation.
But be careful – if you ask the AI for improvements, it will be compelled to provide them. The AI will generate suggestions regardless of text quality, so be very selective about which edits you take to heart.
4. Getting over a mental block
Despite all of us having years of experience writing, writer's block remains a very real issue. We get stuck. A lot.
One thing that helps tremendously is complaining about it to colleagues. Venting your frustration, hearing everyone else's current struggles, and turning it all into a mini brainstorm session tends to get the ball rolling again.
But sometimes you work from home. Other times you don't want to disturb teammates in deep work. That's where AI helps – it can be anything from asking it to finish a sentence to describing the issue you face and requesting solutions.
5. Generating alternatives
Writing is often about condensing key information into a particular narrative. For any specific article, this can be done in dozens of different ways and it's not always immediately clear which is the best approach.
What content to include? In what style? In what order? In what tone? It's common to start writing and realize something's off and then pivoting or starting anew.
With AI, you can quickly generate multiple ideas and see how it all comes together.
Personally, one way I frequently use ChatGPT is to generate multiple alternatives to article and blog post titles. The H1 is one of the most important elements of the piece so getting it right is imperative.
In practice, the AI's title suggestions tend to be terrible. But seeing a variety of options – differing in style, tone, wording, length, clickbait-ness – is great help in constructing a catchy final version.
6. Finding synonyms & better terms
A few years ago, I had a dedicated tab for Thesaurus.com that I'd never close. Today, it has been replaced with ChatGPT – generative AI tools are perfect for finding synonyms and antonyms, and so much more.
You know that feeling when you know a word exists, but you can't recall it? Describe it to the AI and it'll serve it right up. Is there an idiom or turn of phrase that has slipped your mind or may be better suited in a particular context? Just ask the AI.
AI is like a thesaurus on steroids.
7. Transcription
One of the areas of our work where the impact of AI is most noticeable is transcription. We conduct a lot of interviews, client calls, and meetings, and not a day goes by without a call or two or three or four… (help).
That's why we've been using transcription tools like Otter.ai to stay organized. By transcribing audio content, you can quickly find important information without having to revisit and fast-forward through the audio recording.
Thanks to the integration of AI by the transcription tools, the quality of the transcripts has improved dramatically, further simplifying our day-to-day work.
Only half of content creation is writing
It's inevitable that some of the issues current AI models suffer from will be remedied in future evolutions. Artificial intelligence will grow more capable and there may come a time when the writing itself will be done entirely by machines – and done very well.
But content is just as much about writing, as it is about style, composition, representation, emotion, empathy, brand voice, and more. And even if AI's content production quality improves, for the writing to be valuable for your specific business and your specific audience, the AI will still require the guiding hand of an expert.
Ours 👇
Get in touch at hello@truesix.co!
May 9, 2024
ByViesturs Abelis
In