History Function to avoid AI claims

I participate in an art group and submitted my writing, written in scrivener and pasted into deviantart (not discussing the downsides of dA here, please), and my group ran my writing through an AI detector and claimed AI on 2 of my writings that I personally wrote myself, with absolutely no aid at any stage from AI. I fully created it on my own!

But the group would not believe it, because their fancy paid program said “No, this is 97.X% likely heavily assisted or fully written by AI” and I had NO way of proving otherwise because Scrivener does not possess the same sort of “history” function that google docs hosts, which will show the writing as it goes along and auto saves, with nothing needing to be done extra by the writer…it just logs the history, which is apparently all the proof this group needs/wants.

Would it be possible to get a feature like this?

Is a feature like this in development at all?

Anything to prove incremental work was done ONLY by the user and NO AI whatsoever was used at any stage to help the writing?

I REALLY love Scrivener for my creative writing, and would love to ONLY use Scriv when I want to write, but at present if I want to participate in my hobby writing things to submit to this art group I’m in, I can’t use Scriv until there is some sort of way to prove I wrote the work…which as it stands, the program does not seem to possess any way to show incremental work, not just word/character/etc counts, on a given piece.

I’ll probably still use scriv for other things, but this has been a major blow to my steam on my writing journeys.

Thanks in advance for any hope / leads anyone may provide…

if there is an email I can write to the developers on, maybe that would be an avenue to request such a feature if it does not exist yet?

Much Love,

-Melo

Hi.

You could simply link the snapshot function to manual saves…

3 Likes

I think you need to find another art group, honestly. “AI detectors” are notoriously inaccurate. Worse, IMO a critique group that insists on using them is probably not going to develop the trust that honest feedback requires.

To the substance of your question, yes, snapshots are the way to go.

6 Likes

How would that even work? You could have used ChatGPT to create a story and simply type that into Scrivener with this imaginary “prove” function.

As @kewms said, these “AI” detectors are known to create lots of false positives and can’t be relied upon. I, too, think the solution cannot be a technical one but a social one. Find other peers.

7 Likes

What @dirkhaun said.

Same thing applies to the Google docs history feature. (Or any history feature.)

Day 1 - I use AI to write a portion of my paper, then paste the results into Google docs

Day 2 - I use AI to write more of my paper, then paste the results into Google docs

Etc.

Day N - I present my paper to the group, with Google docs history to “prove” my paper is AI-free.

Your art group needs to think this through a little more.

Best,
Jim

4 Likes

The fact that this is good enough for them, along with their trusting expensive scam software that claims to “detect” this stuff in the first place (which you now know first hand is 97% garbage) is proof enough of them being misguided.

I agree with the rest, they are doing you a favour: exit stage right and find a better group. The problem here isn’t the software.

It would be a trivial matter to make a setup that causes an LLM (or if not that, a block of text generated by one on your clipboard) to “type” it slowly (at least for a computer), with minor variations between keypresses, into a webpage (or any other software) like a person would. My twenty year old AlphaSmart almost does that as a lo-fi way of getting writings off of the device and into a computer. You plug it in like a keyboard, press send, and it “types” into Scrivener (or Google Docs, fooling it into verifying the vintage keyboard’s humanity).

Or in other words, Scrivener doesn’t need a fancy marketed “proof of human origin” feature because no such thing exists, and we aren’t disingenuous enough to pretend it does. Unless you’ve got a group of people breathing down your neck behind you, watching you type, and verifying the keys you press are the same ones creating text on the screen in front of you and that what appears on the screen stops whenever they randomly tell you to pause, all of this is wand waving, hyperventilating and parting people from their money (or their data, in the case of Docs, which equals money to Google).

4 Likes

That’s going to be an epic want ad on the bulletin board.

3 Likes

Rule 34 scares me into not searching too deeply into that one.

3 Likes