Beware the happy valley: why unsophisticated AI written copy is easy to spot

As we all see more AI-produced content, it’s easier to spot. It can annoy people with “#AISlop” prevalent on social channels. 

Our concern is that when people identify AI content it sends a subconscious message that not enough care has been taken. It follows that any sense that audiences have been taken for granted will negatively impact their brand perceptions.

Here’s our thoughts on the tell tale signs which will signal you have used AI for your copy, and which you need to ideally, erase:

Hyperbolic language – AI loves to over-emphasise. Even small trends can be a ‘paradigm-shift’, while minor points are often “crucial,” or “imperative. Why does this happen? Well a human writer who finds something genuinely interesting can just say why — they have a specific experience or perspective to point to. AI doesn't have that, so it signals importance through volume instead. If it can't tell you why something matters, it just turns up the adjective dial. "Significant" becomes "groundbreaking." "Useful" becomes "transformative." The overuse of adjectives is a big give away and is to be avoided. Prompt the tool to pick them out or edit them yourself.

The power of three – AI loves a list of three things. Three is the smallest number required to create a pattern and it also feels quite rhythmic when reading. Think of "Life, liberty, and the pursuit of happiness" or, "Location, location, location,” or, “Eat. Pray. Love.” You get the drift. As AI is trained on the sum of all human writing it has internalized this structure as the standard for clear communication. And repeats it over and over. So switch to longer lists, or a narrative flow which avoids the use of bullets, and especially bullets of three!

The cliché – (in “today’s competitive world”), AI is trained on a lot of business communications which tends to use this kind of scene-setting statement, so it often averages that style back out at you. It also avoids committing to anything (remember AI doesn’t have points of view) and is vague enough to fit almost any topic which is exactly why models default to it. The fix is simple, just find the clichés and cut them. Better still, when writing context for pieces, always add your own point-of-view or experience in. By default it will feel more human. And then work with your preferred AI tool to write the meat of the piece, prompting and cajoling to get to copy which feels natural and in your tone of voice.

Signature vocabulary – certain words spiked after 2023 in ways that have nothing to do with how language evolves. "Delve," "tapestry," "it's worth noting," "multifaceted"...These cluster together like red flags. Where there's one, there are usually others nearby. Wikipedia has documented the trend. It came about because ChatGPT was trained on a large amount of academic content which used these kinds of words. It then repeated them in responses to prompts and requests which created more content with the same words and effectively, an echo chamber. 

Academic copy is usually pretty dull and it just feels off when you read it. Keep an eye out on Wikipedia too and delete words which appear in its rogue’s gallery of overused words.

Structural sins – AI has formatting habits which are hard to miss once you're looking. Em dashes appear constantly. Headings end up In Title Case For No Real Reason. Bold text gets scattered around mechanically. There's also a specific inline-header list format that's almost exclusively AI-generated, where each bullet starts with a bolded label followed by a colon. Again, just look for them and smooth them out.

A lack of simple verbs – AI tries to sound smart even when simple is better, so it swaps out "is" and "are" for more elaborate constructions. "The company is profitable" becomes "the company continues to demonstrate strong financial performance." Studies found a 10%+ drop in usage of basic linking verbs in written content after 2023 – largely thanks to ChatGPT.

Vague endings – real writers end with something thought provoking or conclusive. AI wraps up with lines like "The future looks bright" or "Exciting times lie ahead as they continue their journey toward excellence." It gestures at significance without actually saying anything. 

Here’s what to do

Check out our blog on how to write well, and train your preferred AI tool to follow those rules for you. Also, put your spin on everything, using personal insights and experience. This is our most important tip because your point of view is unique to you. When you blend the human and AI in this way you get all the benefits of AI in terms of accelerating your copy while producing work which feels like it’s truly yours. 

Also, methodically seek out and remove common AI “tics” – especially the rule of three, clichés, the adjective blizzard, em-dashes and overused words. It’s always useful to come back to your work a day or so later as well before publishing it. Checking it with a fresh eye will always reveal areas for improvement.