AI Checkers: A Copywriter’s Perspective

With the explosive rise of ChatGPT and Bard, I have noticed that many job posts have included a phrase like ‘No AI allowed, all submitted work will be checked!’

While I understand the concern of clients and I, in fact, very much share it,  I also would like to warn of an over-reliance on AI checkers as a holy grail for solving this problem. It is quite strange for me to see that we all collectively (and rightfully) doubt the facts and figures AI throws our way but will blindly trust checkers to confirm if a text is written by a human or not. To illustrate my point, I ran an article I recently wrote for iGaming.com about Jan Pieterszoon Sweelinck (a Dutch composer) through the four highest-ranked results on Google for ‘AI checker’.*

The results are as follows:

ZeroGPT – 33.09% AI-generated, classified as ‘Most Likely AI/GPT generated.’

Contentatscale.ai – 0% AI-generated, classified as ‘Highly likely to be Human!’

Copyleaks.com – 92,3% AI-generated, classified as ‘AI Contented Detected.’

Contentdetector.ai – 0,6% AI-generated, classified as ‘Likely Human Content.’

As you can see, only two out of the four realised that I entirely wrote my text, and one of them did not even think it was written by a human at all. Interestingly, both ZeroGPT and Contentdetector agree that I do not fully write my text but wildly disagree on the percentage of the work that is mine. 

While AI checkers can be useful as a tool in the fight against content by bots, I would like to stress the importance of not believing them to be the infallible arbiters of authenticity. Let’s have a little faith in copywriters, as I firmly believe their creativity and originality are still far more brilliant than what AI can produce.

*This excludes Writer.com, as they only check the first 1.500 characters, and my article is longer.