Pangram

“WARNING: If you're getting an unexpected result you may not be on the actual pangram site. My search result listed a "sponsored" Pangram site which appears to deliberately produce incorrect results. The URL of this bogus site is textguard.ai. It told me that my text (below) was 77% AI generated. I later realized that this was a nefarious site. The Pangram site corrected identified my text as human-generated. I've spent years working on an idea introduced by an essay. Pangram correctly sa”

“I wrote this essay myself, about my life and English is my second language. AI corrected my grammar, that is all. Pangram told me it was '100% AI, the entire text.' That is seriously BS. Human-written with grammar help is not '100% AI,' and a tool that cannot tell the difference has no business being that confident about a person's work. Thank you for ruining my valuable work that was called out by idiot people who deeply trusted in your silly AI who cannot tell the difference. SHAME ON YOU!!! ”

“This is the only AI classifier I’ve used that works. Any time I’m reading a blog post or article and I have a suspicion that it is AI generated, I check it with Pangram. They warn you up front that it can give inaccurate results for passages under 150 words, but for longer text, it’s pretty much a guaranteed accurate result. That said, I cannot prove it, BUT I am willing to bet that the people who spam this product with 1 star reviews and claim “It flagged something as AI that I wrote MYSELF” di”

“This is not consistent and unreliable. This flagged my cover pages/title pages/index at 88% AI, and marked my reference page as being plagiarism. Also indicated a literature review that I did, which has in-text citations throughout, as AI. It's cited because they aren't my ideas/words; credit was given to the original author, and it is not AI. This also flagged the data I collected from a survey as being AI, when it is actual data points that I received, and I attached an image as an appendix”

“Pangram Seems to Work and Students Don’t Like It As the fall semester approaches I have two questions. How good is Pangram.com at avoiding false positives for AI prose, and has student AI use grown since ChatGPT's arrival? Fortunately, I have a dataset with which to answer these questions: seven years of reading responses from my Communication in the Digital Age course. Unfortunately, I don't have all the students' responses over that time, but many of their responses do survive on the students'”

“This is not consistent and unreliable. This flagged my cover pages/title pages/index at 88% AI, and marked my reference page as being plagiarism. Also indicated a literature review that I did, which has in-text citations throughout, as AI. It's cited because they aren't my ideas/words; credit was given to the original author, and it is not AI. This also flagged the data I collected from a survey as being AI, when it is actual data points that I received, and I attached an image as an appendix”

“I am being forced to use this AI detected tool for courses at Capella University. This tool give false positive results. I have screenshots of proof showing that when I put a single paragraph into the tool that was completely written by me without any AI help, I get the result of 100% human written. Now if I put all the paragraphs together into this tool, the result is 100% AI written. How can the same individual paragraphs submitted go from 100% human written to all paragraphs submitted tog”

“Pangram AI tespit programı, metinler üzerinde tutarlı ve güvenilir sonuçlar vermemektedir. Aynı metne yalnızca bir cümle eklediğimde oranı %0’dan %100’e çıkarması, uzun bir metne eklediğimde ise oranı tamamen farklı değerlendirmesi programın ciddi biçimde hatalı çalıştığını göstermektedir. Ayrıca hiçbir değişiklik yapılmadan tekrar test edildiğinde bile sonuçlar sürekli değişmektedir. Teknik ifadeler, sayısal veriler veya insan tarafından yazılmış nötr cümleler dahi %100 AI olarak işaretlenebilm”

“Extremely easy to use, and it makes detecting AI generated content or partially AI content easy to spot.”

“Great review performance, but still cumbersome interface and capacity to streamline results. Also unable to handle large files with a lot of images”

“So at my first opportunity, I decided to check whether or not I could permanently disable the "AI detection" fraudulent tool on Substack publications and there's no way to actually do it without actually using the tool you want to disable. According to the "Substack Support AI", the only way to disable the fraudulent Pangram detection AI is to first run the AI on your posts and to individually then disable the AI detection tool per post. But you can only do this AFTER you've run the "AI detectio”

“Just FYI, you can turn it OFF and REPORT false positives. But you can’t do it until it’s already scanned your texts (that’s sus in itself). I think if enough people turn it off and report false positives maybe Substack will rethink their decision to partner with this flawed company before all the good writers leave.”