Suspiciously few people care enough to notice such things and then put 2 and 2 together.
The only explanation I have, has to do with the suspiciously numerous people who care suspiciously much about the necessity of calling you names when you do identify some glaring contradiction, security hole, or the like.
"How dare you perceive what is right in front of your eyes! You must instead perceive the imaginary things that everyone is talking about (or, shh, be destroyed, hehe)" is a common enough token sequence that human languages usually have got it compressed down to 1-2 words.
Al Chatbots officially allowed for all aspects of Linux Kernel development. “Kernel contributors have been using [Al tools] to generate contributions for a long time.”
53 days after Cloudflare announced that their service was now “more secure” because they had ported it to Rust... an error in their Rust code caused a massive outage.
The Em Dash ( — ) has been in use for centuries by some of the most renowned writers — from journalists to classic literature to modern Sci-Fi authors. Not a good way to identify AI output.
I would say that in the realms of stylometry, unless it's visible in a corpus from an author before AI emerged, -- it would be very plausible to say for that author, the use of em-dash was indicative of having copy-pasted from the generation of AI systems which used it as a marker.
After this is all history, it won't be a good determinant.
If enough people begin using em dashes in some coherent manner, again in the realms of stylometry, it will be a statistical statement if their pattern matches the ones humans use, specifically this human, or the ones AI use, specifically these AI.
These kinds of things (it turns out) can be quite hard to fake. It's like when people try to act as a random number generator, the patterns they emit simply don't act like the RNG. I doubt people using em dash will match how robots do viewed at a distance, in volume.
I feel confident that if Jane Austen or Samuel Johnson use em dashes, (or their printer at least) we can state with some confidence, it's not an AI. IF however comparison of the ASCII from Gutenberg, and an apparent "AI cleaned up" copy of their work diverge, what would we make of things like this? Is it material change? For "Tristram Shandy" it would be: Sterne deliberately told his printer to put some things into the text, such as a black page, and some other printer marks. If they get elided out, or magnified up, thats dicking with the text.
Two weeks ago, The Lunduke Journal pointed out how many Leftist Open Source organizations (Rust, NixOS, Linux Kernel, etc.) still used the “Master” branch naming (against their “inclusive naming” rules). Yesterday, Rust changed to “Main”.
> Why were web sites ever able to access your local network without your permission? This seems like it was always a bad idea.
reply