Questions to Copilot about AI Detectors
This is me asking AI Copilot a few questions about AI detectors.
Me: What AI detector program would you recommend for teachers?
Copilot: There are several AI detector programs that teachers find useful for maintaining academic integrity. Here are a few highly recommended ones:
Winston AI: Known for its accuracy, Winston AI is a popular choice among educators
CopyLeaks: This tool is praised for its ability to detect AI-generated content effectively
Content at Scale: Another reliable option that helps in identifying AI-written text
AI Detector Pro: Offers comprehensive reports and is highly rated by teachers
Scribbr: An all-in-one tool that also includes plagiarism detection
Turnitin: A well-known tool that has been updated to detect AI-generated content as well.
Each of these tools has its own strengths, so it might be worth trying a few to see which one best fits your needs. Do you have any specific requirements or preferences for an AI detector?
ME: I tried Scribbr and it seemed fine. Would an AI detector use keywords in determining its usage?
Copilot: Scribbr is solid! AI detectors like it use various techniques, including keyword analysis, to identify AI-generated content. They often look for patterns in language use, sentence structure, and even stylistic elements that are typical of AI writing. Some might also use machine learning models trained to detect AI-generated text based on these factors. It's not just about keywords, but the overall "feel" of the writing that can give it away.
ME: Would it be ordinary for an AI detector to say something was generated by AI because of the use of the phrase vivid imagery?
Copilot: It wouldn't be typical for an AI detector to flag a text as AI-generated based on a single phrase like "vivid imagery." They analyze the overall pattern of the text, such as the consistency, complexity, and unique identifiers of human vs. AI writing. While certain phrases might be more common in AI-generated content, reliable detectors look at the bigger picture to make an accurate determination.
****
Other than my questions everything else was written by Microsoft Copilot.
karenina
Poet Rating
Short Works Rating
Review Stars
Rank: 27
RE: Questions to Copilot about AI Detectors
Great insight into a growing problem and imperfect solutions, Gloria.
None of us support AI generated posts or reviews, and I can attest to the frustration when being falsely flagged because of the use of "computer-like" words.
(Whatever those are)
I received a reply from Tom upon receiving my latest "warning" that I should disregard, because "they" know I am not generating AI reviews.
I know that as well!
No one wants to be in a data base as having been (falsely) accused of infringement of site rules, and there is no way to "prove" we have not done so!
I hope the site fine tunes their AI detection project.
nor84
Level 4 Pro
Poet Rating
Short Works Rating
Novel Rating
Review Stars
Rank: 237
RE: Questions to Copilot about AI Detectors
AI is a bit like the atomic bomb. Somebody invented it as a necessity, and now we're stuck with it. It has taken over the audiobook industry, the Internet, and the job of receptionists in many businesses. It isn't going away, no matter how annoying it becomes.
SimianSavant
Poet Rating
Rank: 16
Short Works Rating
Novel Rating
Review Stars
Rank: 117
RE: Questions to Copilot about AI Detectors
Message edited:
AI has not "taken over the Internet". It has definitely taken over the majority of images used at Fanstory though. A few months ago Gloria and I were among the very few using it (I think she beat me to the punch). We will increasingly find however that it pervades, amplifies and dominates content in many more areas and that it becomes difficult to compete without using it, similar to previous disruptive technologies, and it will massively accelerate and evolve in dramatic ways unexpected.
Gloria, did you try running the AI responses from Copilot through an AI detector?
RE: Questions to Copilot about AI Detectors
After your question H, I did run the Copilot responses through Scribbr and the result was that 100% of text is likely AI-generated.
That's a rather ominous prediction. If I were to go Sci-Fi, my guess is that AI would attempt to hide itself from the detectors. But so far they seem to be effective.
SimianSavant
Poet Rating
Rank: 16
Short Works Rating
Novel Rating
Review Stars
Rank: 117
RE: Questions to Copilot about AI Detectors
A possible ironic benefit of AI is that it may teach humans to write less generically, carefully and stupidly in exactly the way that AI with its woke HR masters have forced it to be.
RE: Questions to Copilot about AI Detectors
A fine and helpful post. My words to that effect were not taken seriously, then ignored. I'm glad to see that most people think this particular AI "detector" is incompetent, including the AI detector Copilot. Ironic.
