Post

yes, I checked since thanks to our tech overlords you have to do that these days - anyway this guy has been on reddit for 13 years and has a pretty consistent voice - so no, I do not think this is AI

and has referenced a number of specific military experiences over a number of years

2 reposts422 likes

Does it matter here. Should he use a quill and a dictionary for spelling to make it legit? I presume he read it before posting but he certainly posed the correct questions. To address the tech issue, it's a program, not smart. Don't let the tech bros, like the T Admin does, distort your perceptions.

It’s still AI-written; he prompted the AI to try to ape his manner of speaking, and perhaps he did some post-generation editing, but it’s still entirely obvious. Go to his profile and read some comments from 7 or 8 years ago. They are quite different in style.

it could be that he was really young 7 or 8 yrs ago (lots of redditors make accounts as teens) and got better at expressing himself on the internet as he aged. just looking at his comments over the past 2-3 years nothing strikes me as inconsistent. A ChatGPT version of this would be flatter imo? 🧐

It seems like it might have been his own sentiments run through AI for polish - and yes, as I specifically said, I looked at his old posts.

I’ve certainly heard of a lot of normal people doing that these days - running a LLM over their own writing to smooth it out. I don’t love the practice but it’s also very arguably not inorganic either.

I predict that in the future everybody who doesn't post under their real name (which will also become a larger % of all posters) will run their text through an LLM because it seems like it'd be harder to dox someone whose posting style just looks like every other LLM user.

I find it impossible to write a self-review for work so I let an LLM expand on my sentences. My company is pushing it on us so I figured that was a self-review was a good medium to use it.