    |
|
|
|
|
|
|
Artificial intelligence (AI) is increasingly being used as a tool for writing texts, for a wide variety of purposes.
The use of AI can lead to conflicts when someone claims to have written a text that was actually produced by artificial intelligence. There are various digital tools available to detect AI usage in a text, some of which are free. The issue when using them is: are these tools actually reliable?
I have been writing this blog for 22 years. I enjoy writing, and I don't want AI to take that pleasure away from me. I decided to test several of these verification tools using a text I wrote years ago—specifically, an article from November 2018 about Spain and Poland. When I wrote it nearly eight years ago, the AI text-generation tools we have today did not exist (or, if they did, they certainly weren't available to the general public). You can verify the text's publication date, as it was archived on the Wayback Machine in 2018. If I had made any changes, they would be visible there.
Since not all verification tools can process large amounts of text, I ran the check using the article's first three paragraphs (those appearing before the first photo embedded in the text). This amounts to a total of 2,187 characters. Here are the results I obtained from various verification websites:
It is interesting to find such disparate results. While some identified the text as 100% human-written (which is correct, and I can vouch for that), other checkers seemed to think it was almost entirely written by AI, when that is not the case. In any event, this test helps me determine which checkers to use should the need arise, but above all, it shows that AI—since those tools ultimately run on artificial intelligence—is not as infallible as many people believe.
---
Image: Meta.ai.
|
Don't miss the news and content that interest you. Join Counting Stars for free on Telegram: Click here to join |
Opina sobre esta entrada: