Hi Tomasz! Yes, however that's just a vacation from Traffic Research so I can catch up on other things that I need to do. I really need to take a full vacation as well :| There's so much going on right now that it's hard to stay away. Hope you're doing well!
Yes and they do. For a simple test (outlined in Google's most recent AI detection research paper), they ran a sample 500 million pages. This isn't a hyperbole. They ACTUALLY used 500 million pages as sample data. So it goes to reason that if they can process 500 million pages as sample data, they could easily (and have) rolled this out for the web once they confirm the data aligns with their objectives.
This passes industry standard AI detection and is guaranteed to pass our roBERTa based detector (same as Google used in their research paper). Unfortunately GPT-zero also thinks the Declaration of Independance was AI written. GPT-Zero thinks the articles my writers wrote back in 2013 were AI written. As great as the story of a kid coding GPT-Zero in a weekend sounds, unfortunately it's not reliable and produces too many false positives. It can be considered a fun experiment but shouldn't be used as a legitimate way of detecting AI. I sympathize for the kids in school that are being falsely accused by their professor of using AI to write their essay because the teacher mistakenly used GPT-zero to check it. This is after spending close to 100+ hours working on AI detection, including developing and rolling out our own AI detection service.
@Traffic Research Yes I did notice the flaws you mentioned, it also can be fooled by randomly using 2 spaces instead of 1. Could you please refer me to a free ai detector that is more accurate?
I own an SEO Agency and am interested in this tool but $229/month for 10 articles is a little crazy. Am I reading that right?
Hi Eric, I thought you are in vacation mode
Hi Tomasz! Yes, however that's just a vacation from Traffic Research so I can catch up on other things that I need to do. I really need to take a full vacation as well :| There's so much going on right now that it's hard to stay away. Hope you're doing well!
price?
Can G afford to check trillions of pages to detect AI content?
Yes and they do. For a simple test (outlined in Google's most recent AI detection research paper), they ran a sample 500 million pages. This isn't a hyperbole. They ACTUALLY used 500 million pages as sample data. So it goes to reason that if they can process 500 million pages as sample data, they could easily (and have) rolled this out for the web once they confirm the data aligns with their objectives.
Bro this doesn't pass gpt zero
This passes industry standard AI detection and is guaranteed to pass our roBERTa based detector (same as Google used in their research paper). Unfortunately GPT-zero also thinks the Declaration of Independance was AI written. GPT-Zero thinks the articles my writers wrote back in 2013 were AI written. As great as the story of a kid coding GPT-Zero in a weekend sounds, unfortunately it's not reliable and produces too many false positives. It can be considered a fun experiment but shouldn't be used as a legitimate way of detecting AI. I sympathize for the kids in school that are being falsely accused by their professor of using AI to write their essay because the teacher mistakenly used GPT-zero to check it. This is after spending close to 100+ hours working on AI detection, including developing and rolling out our own AI detection service.
@Traffic Research Yes I did notice the flaws you mentioned, it also can be fooled by randomly using 2 spaces instead of 1. Could you please refer me to a free ai detector that is more accurate?