Will Apple Intelligence be able to warn me when AI has been used to write something?

It's the only case I want AI for.

Posted on Nov 25, 2024 08:31 AM

Reply
Question marked as Top-ranking reply

Posted on Nov 25, 2024 08:37 AM

If you are asking it you can use AI to tell you whether an article, text or book, was generated using AI, no.

7 replies

Dec 3, 2024 09:38 AM in response to noit x

noit x wrote:

It's the only case I want AI for.


Detection of LLM generated text is not (reliably) available so far, though some of the efforts are getting less-wrong.


https://www.zdnet.com/article/i-tested-9-ai-content-detectors-and-these-2-correctly-identified-ai-text-every-time/


I'd expect legal entanglements to arise for the folks using LLM-based detection, and particularly around false positives, though quite possibly also around false negatives.


LLM-generated postings around here tend to have pretty formatting, bolded text section headers, perfect spelling, and variously include irrelevant or outdated info. They do tend to be easier to spot. At least so far. But I could be AI, after all.


LLM-generated text is entirely statistical, which means it's effectively guessing at what the next word in the generated text should be, based on the prompt and on a whole lot of text previously scraped and ingested into the model. Input text which can be correct, or can be incorrect. Detection is the same thing: a statistical guess.

Nov 25, 2024 11:25 AM in response to noit x

There are AI detectors out there. However, their accuracy is very questionable. One of the first ones stopped offering the service because the developer said it just wasn't accurate enough.


Here are a couple an internet search turned up:


https://www.zerogpt.com/

https://www.grammarly.com/ai-detector


Note that I haven't used either of these and don't know much about them. I can tell you that I use Grammarly's proofreading product and like it.


Funnily enough, one of the sites that popped up advertised itself as a tool that would tell you if your AI-generated content would be flagged as such and, if so, how to "humanize it."

This thread has been closed by the system or the community team. You may vote for any posts you find helpful, or search the Community for additional answers.

Will Apple Intelligence be able to warn me when AI has been used to write something?

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.