pabloblanco
Newbie
- Aug 8, 2023
- 7
- 5
As part of the AI content work I've been doing for my journey I've been giving some thought to the concept of detecting AI content and I wanted to throw my ideas out there to see what others in the community think.
In the future there will only be good and bad content.
I believe that how text is generated won't matter as much as the accuracy of the article and the authority of the author.
https://decrypt.co/149826/openai-quietly-shutters-its-ai-detection-tool shows the challenges of being able to correctly classify content as written by AI and more importantly not giving a false positive to human created content.
I spent a few hours playing around with some of the other classifier tools, using content from case studies published by leaders in our field and sure enough the tools did flag many of these articles as AI to some degree. I tried to remove the chance of these being AI generated by using archive.org to look at the site pre-2019.
While Google could build a massive training set of AI articles and non-AI articles I believe that this might be too massive a challenge even for Google and probably not worth the effort.
From a useability point of view, it makes sense to serve the best possible result to the user. So if we have a poorly written article by a human and a well written article by an AI, which should they rank? For me the answer is obvious.
Anyway that's my little rant about the state of AI. Thoughts welcome.
In the future there will only be good and bad content.
I believe that how text is generated won't matter as much as the accuracy of the article and the authority of the author.
https://decrypt.co/149826/openai-quietly-shutters-its-ai-detection-tool shows the challenges of being able to correctly classify content as written by AI and more importantly not giving a false positive to human created content.
I spent a few hours playing around with some of the other classifier tools, using content from case studies published by leaders in our field and sure enough the tools did flag many of these articles as AI to some degree. I tried to remove the chance of these being AI generated by using archive.org to look at the site pre-2019.
While Google could build a massive training set of AI articles and non-AI articles I believe that this might be too massive a challenge even for Google and probably not worth the effort.
From a useability point of view, it makes sense to serve the best possible result to the user. So if we have a poorly written article by a human and a well written article by an AI, which should they rank? For me the answer is obvious.
Anyway that's my little rant about the state of AI. Thoughts welcome.