5 easy ways to tell if your written content came from GenAI

It should come as no surprise to anyone that we are seeing more and more AI-generated written content online as more genAI platforms are released and older models are upgraded. But it can be shocking to learn how quickly the pace of AI content is being added.

According to research reported by AI detection firm Copyleaks earlier this year, the growth rate of AI content from the release of ChatGPT 3.5 in November 2022 to March 2024 was a staggering 8,362%. Granted, the initial base was very low, but over this time frame the numbers are telling. The findings show that:

• The volume of AI-generated content on the internet increased by 187% from November 2022 to January 2023.
• Over the next 12 months, the amount of AI-written copy increased by 2,848% across the web.
• While a McKinsey study from March found that the use of genAI jumped to 65% of respondents for 2024 from 33% in 2023.

And there’s no sign of genAI usage slowing down. Now more than ever it is important to be able to distinguish between AI content and human-generated content for the sake of digital integrity.

Here are five ways to tell if your written content came from an AI bot or not.

Language Patterns

You can often see in AI-generated text that it lacks emotional subtlety and tends to use overly formal or complex words. So the sentences sound stiff or boring without the grain of quality human writing.

AI sentences also do not vary in length. Perhaps due to their training or programming, their sentences are eerily similar in word count. They can read very monotonously and keep running.

Consistency issues

The AI ​​also seems to struggle a bit with consistency when it comes to narrative details, especially when trying to write fiction. You may notice that it has abrupt changes in the description of a setting where a scene occurs without explanation or details of a character appearing without warning.

Or there may be inconsistencies that don’t line up with the previously mentioned events throughout the story that show the limited understanding of AI contexts at times. Regardless, it’s something to watch out for.

AI tool detection recommendations

Copyleaks was mentioned at the beginning of this write-up and claims on its website that it is the most accurate AI detector based on multiple independent studies. One feature that makes it stand out from other solutions is that it works in 30 different languages.

Another widely used AI detector for writing is GPTZero. It is very effective at analyzing text patterns, overused phrases, and common word combinations that GPTs tend to pass. It offers a free version up to a certain character limit, with more features for the paid options.

Below are two basic scans of an uploaded submission written by a human from 2021 before the advent of genAI – the screen correctly marked it as human production.

The second image was a poem written with ChatGPT that was uploaded today and was also pinpointed. Both Copyleaks and GPTZero are useful, and it’s best practice to use more than one.

Depth and context

Another key giveaway in AI writing can be a lack of detail, vagueness, or overgeneralizations. For some reason, when they write their AI models, they tend to avoid deep contextual or strong topical understanding.

The current generation of these productive pre-trained transformers revolves around complex subjects that require personal experience or specialized knowledge. The explanations provided may be delivered confidently, but without supporting details to validly support the point.

Unusual language errors

Although rare, AI-generated texts sometimes include odd mistakes or turns of phrase that are awkwardly out of place. An example I personally saw was a sample AI article on climate change where trees were once referred to as “Arbor Day notables.”

Such unusual combinations of common phrases, slurs or basic grammatical errors – such as the recent dust up over ChatGPT claiming that there are only two letters “r” in strawberry – are useful hints that the text comes from artificial intelligence.

It is clear that we are in the early pages of the genAI time book. Since AI isn’t going anywhere—except to new levels of rapid adoption—it’s our responsibility to keep up with the latest developments in both AI creation and AI detection so that AI isn’t left to write its own story.