1.9 C
Munich

How to find out if a text was written by AI: First steps for Google too

Must read

We started 2023 by chatting with ChatGPT, the artificial intelligence of OpenAI. Something told us that this will be discussed a lot soon, and indeed it was: Between Microsoft’s maximum possible investments and more, the Tech world’s eyes were focused there. But How to recognize AI text? Let’s examine the subject in depth.

Also, thanks to the increasingly persistent rumors that Satya Nadella and her partners want to integrate ChatGPT into the Bing search engine (and not only), in early 2023 Potential “invasion” of AI-written content on Google And so on. In fact, understand well that this may be a non-trivial question that users are starting to wonder about too.

It is worth noting that, according to the current situation, ChatGPT was discontinued in 2021. In fact, the historical information of the AI ​​database stops in that year, as ChatGPT was not able to connect to the internet. Simply put, OpenAI’s solution cannot write up-to-date content like today’s news because the response in these contexts is “As my training stopped in 2021 and I do not have access to real-time sources of information, I cannot inform you about current or future events.“.

Potential problems that already existHowever, they are also linked to content that stays evergreen, as you well understand that search engines have a little bit of everything. Also, from a future perspective, this type of AI-written content and text that was initially “copied” from other sources and then “rewritten” by AI could really start to proliferate. In this context, the main players in the industry are currently trying to take cover.

For example, as pointed out by Search Engine Log, it refers to: AI content watermarkor adding a “watermark” that allows to identify everything. OpenAI seems to be working on it, even if the discussion about it is not trivial and includes potential methods that could be used by bad actors to “bypass” the watermark. For example, one potential method is to have another AI pass through to “interpret” the first output.

In short, we begin to think about everything in an important way. By the way, as mentioned by LinkResearchTools, already find out which texts are written by AI. Among them, the website PoemOfQuotes seems to be among the most interesting (though adding an Italian text taken from the portal ChatGPT, indicating that it was probably written by a human).

But these are not exactly reliable public services and far from them. already available on google. In fact, the latter already seem to be able to detect at least some AI content, making them less relevant among search results. The management of such texts is still in its infancy, but it’s interesting to see how many large companies and OpenAI itself seem willing to “limit” everything for ethical reasons.

Source: Every Eye

More articles

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Latest article

Related