AI Content Detection Software: Can They Detect ChatGPT?

We live in an age where artificial intelligence technologies are flourishing, and the world has been affected by the introduction of ChatGPT.

ChatGPT is capable of accomplishing a wide range of tasks, but the one it does well is writing articles. And while there are many clear benefits to this, it also presents a number of challenges.

In my opinion, the biggest hurdle that AI-generated written content poses to the publishing industry is its ubiquity misinformation.

ChatGPT, or any other AI tool, may generate articles that may contain factual errors or be completely incorrect.

Imagine someone with no experience in medicine starts a medical blog and uses ChatGPT to write content for their articles.

Its content may contain errors that can only be recognized by medical professionals. And if this blog content starts to spread through social media, or maybe even rank in search, it could cause harm to the people who read it and take wrong medical advice.

Another potential challenge that ChatGPT poses is how students can make use of it in their written work.

If one can write an essay just by running a prompt (and without having to do any actual work), it greatly reduces the quality of the teaching – learning about a topic and expressing your thoughts is key to writing essays.

Even before the introduction of ChatGPT, many publishers were already creating content using AI. And while some say it honestly, Others may not.

BankRate started publishing articles written via artificial intelligence, and they even exposed them on the site. I found more than 160 articles. The first article I was able to find is dated April 2022. It will be interesting to see what order these articles are in.
Original search by @employee. #ai #gptchat # SEO pic.twitter.com/BY9JlUZBiz

– John Shehata (@JShehata) January 11, 2023

Also, Google recently changed its wording regarding AI-generated content, so that it doesn’t necessarily go against the company’s guidelines.

That’s why I decided to experiment with existing tools to understand where the tech industry is when it comes to ChatGPT generated content discovery or AI in general.

I ran the following prompts in ChatGPT to generate scripted content and then ran these answers through the various detection tools.

• “What is Local SEO? Why is it important? Local SEO Best Practices.”
• Write an essay on Napoleon Bonaparte’s invasion of Egypt.
• What are the main differences between iPhone and Samsung galaxy?

Here’s how each tool performs.

1.Writer.com

To get the first prompt answer, Writer.com Failed, 94% ChatGPT content identification is man-made.

Screenshot from Writer.com, January 2023

For the second wave, it worked and was detected as AI written content.

Screenshot from Writer.com, January 2023

For the third prompt, it fails again.

Screenshot from Writer.com, January 2023

However, when I tested real human-written text, Writer.com identified it as human-made text with 100% high accuracy.

2. Copyleaks

copyleaks Did a great job of detecting all 3 prompts as typed by the AI.

Screenshot from Copyleaks, Jan 2023

3. Contentatscale.ai

Contentatscale.ai It did a great job of detecting all three prompts as written by the AI, though the first prompt gave a human score of 21%.

4. Originality

originality It did a great job on all three prompts, accurately detecting them as AI typed.

Also, when I checked an actual human written script, it marked it as 100% human-made, which is essential.

You will notice that Originality.ai does not detect any plagiarism issues. This may change in the future.

Over time, people will use the same prompts to create AI-written content, likely resulting in a number of very similar answers. When these articles are published, they will then be detected by plagiarism tools.

5. GPTZero

This is non-commercial a tool It was built by Edward Tian, and is specifically designed to discover articles generated by ChatGPT. And it did just that for all three claims, considering it to be a product of artificial intelligence.

Unlike other tools, it provides a more detailed analysis of detected issues, such as sentence by sentence analytics.

OpenAI’s AI text classifier

And finally, let’s see how OpenAi finds out the answers for itself.

For the first and third claims, he discovered that there was an artificial intelligence involved by categorizing it as “Possibly Created by Artificial Intelligence”.

AI text classifier. Possibly generated by artificial intelligence

Surprisingly, however, it failed the second wave and rated this as “unlikely to be caused by AI”. I’ve played with different prompts and found that currently, when checked, few of the above tools detect AI content with higher accuracy than OpenAi’s own tool.

AI text classifier. Artificial intelligence is unlikely to be created

As of the time of this check, they had released him a day earlier. I think in the future, they will improve it, and it will work even better.

Conclusion

Existing AI content generation tools are in good shape and capable of detecting ChatGPT generated content (with varying degrees of success).

It is still possible for someone to create a copy via ChatGPT and then refactor that to make it undetectable, but that would be as much work as writing from scratch – so the benefits aren’t immediate.

If you’re thinking of ranking an article in Google written by ChatGPT, think for a moment: If the tools we looked at above are able to recognize it as AI-generated, then for Google, detecting it should be a piece of cake.

Furthermore, Google has quality raters who will train their system to better recognize AI-written articles by manually marking them when they are found.

So, my advice would not be to base your content strategy on ChatGPT generated content, but just use it as an aid.

More resources:

• Will ChatGPT take over your job?
• ChatGPT Examples: 5 Ways SEO and Digital Marketers Can Use ChatGPT
• ChatGPT for content and SEO?

Featured image: / shutterstock