Don't Trust AI Detection Devices, This Is The Reason

JAKARTA Technological developments always bring both good and bad sides, including Artificial Intelligence (AI). This technology can increase the efficiency and automation of tasks, but it can also be a means of fraud.

AI is often abused for various purposes, ranging from deceiving lecturers by collecting tasks done by AI, phishing through messages and emails with neat grammar, to creating deepfakes that harm certain people.

With so many types of AI-based fraud, many developers are competing to create tools that are claimed to detect AI with the right accuracy. However, there are several reasons not to believe in AI detection devices.

Wrong Detection Percentage

Reporting from Makeuseof, AI detection devices often show wrong results. For example, human-made content is claimed to be 100 percent made by AI, while content completely created by ChatGPT or other AI models is considered to be 70 to 100 percent man-made.

This happens because human writing and technology are still indistinguishable. As sophisticated as the language recognition method is, this tool can still make mistakes. One of the AI detectors that is proven to have made a mistake is ZeroGPT.

Bible Detection Results

If you want to know the fatal error of the AI detector, you can enter some verses from the alkitab. Our holy code has been made for a long time, even before humans know about technology. If an alkitab verse is detected as AI-made, it is clear that the tool made a mistake.

This error can be found in ZeroGPT. The AI detector showed that the verse of the algorithm was made by AI, a completely false result. This does not only happen to the verse of the Koran, but also to the literature and the text of history.

US Independence Declaration Claimed By AI

The next error shown by the AI detector is the result of the US independence declaration. If you try to enter the text into the AI detector, it will usually claim the text as AI-made.

The text of the declaration of US independence was made in 1776, almost 250 years ago, so it is impossible for the text to be made by AI. Therefore, the analysis of AI detectors cannot be trusted because not a few are still making mistakes.