For what it’s worth I read the whole thing in what felt like one or two minutes and I don’t think I’m a particularly fast reader. I think it looks longer because there are not many blank lines. It seems well written but I guess I do slightly get that AI feeling too, it just might be because he/she is a good writer so now people think good writing is AI, sad it’s coming down to this.
Comment on Self-hosting in 2025 isn't about privacy anymore - it's about building resistance infrastructure
motruck@lemmy.zip 1 day ago
Are all these long form posts written with the help of AI? The length of posts here seem abnormally long for this type of forum. I’m not saying I don’t like it but I’m immediately skeptical when I see a giant post nowadays.
BoycottTwitter@lemmy.zip 1 day ago
Appoxo@lemmy.dbzer0.com 1 day ago
Imagine.
Having to dumb down your writing just because you don’t want users accuse you of being a bot/intelligence agency
rimu@crust.piefed.social 1 day ago
Yes, it is a LLM. Congrats on being one of the very few who noticed.
It even generated “you’re absolutely right” once. Also replied to its own post as if replying to someone else.
PhoenixAlpha@lemmy.ca 1 day ago
Once? Try every single comment before this particular chain, except one. Sure it only generated that exact phrase once, but they’re all variations on you’re right, or that hits hard, or you nailed it, or whatever.
BaconWrappedEnigma@lemmy.nz 1 day ago
This does not look like it was generated by an off-the-shelf LLM. It could be from a custom fine-tuned LLM (or even few shot) but it’s likely not written by vanilla ChatGPT, Gemini, etc…
It can be really difficult to detect LLM written text but the easiest heuristics are:
- Specific keywords
- The use of three examples, often bullet points (Hah!)
- “Final thoughts” or a summary
That said, there are many techniques to make an LLM sound more like an author; so, you never really know…
Final thoughts
In conclusion: we can’t be sure, but at first glance, this looks like it was written by a human.
PhoenixAlpha@lemmy.ca 1 day ago
You forgot one more tell that this post is riddled with - “not x, but y”. The rule of 3 is also seen in general sentence structure as well as bullet points. Example:
A woman was reduced to a data point in a database - threat assessment score, deportation priority level, case number - and then she was killed. Not by some rogue actor, but by a system functioning exactly as designed.
Em-dash (probably), into rule of 3, into em-dash, into not x but y. That sentence is what made me suspicious but there are plenty of other examples.
Well, that and…this killing had nothing to do with any of those points. The sentence sounds flashy but is completely wrong on closer examination. Almost like a…hallucination…ahem.
h333d@lemmy.world 1 day ago
@PhoenixAlpha I’ll be sure to tell my 10th-grade English teacher that her lessons on rhetorical devices are now considered hallucinations. If “not X, but Y” makes me a bot, then half the op-ed columnists in history are running on silicon.
As for the Renee Good shooting, if you think the infrastructure of surveillance, license plate readers, and cross-referenced databases “had nothing to do” with how ICE operates in a city like Minneapolis, then you’re missing the forest for the trees. I’m not here to win a Turing test; I’m here because I’m tired of seeing tech used as a weapon, you know?
PhoenixAlpha@lemmy.ca 1 day ago
The original hallucination:
threat assessment score, deportation priority level, case number
The new hallucination (also rule of 3):
surveillance, license plate readers, and cross-referenced databases
“Surveillance” and “databases” (what does cross-referenced even mean or add? LLMs like to output word salad) could be applicable, but only because they’re so damn vague. Yes, of course the government uses SQL.
License plate readers, sure they were involved…except that wasn’t even one of the original points. Find a model with better context length…lol. They also have nothing to do with self-hosting. What are you gonna do, run your own license plate issuing server?
Please, you can just say you used an LLM because English isn’t your first language or something. I’m literally giving you an out. It would be way less embarrassing than whatever you’re trying to accomplish.
someone@lemmy.today 1 day ago
shut the fuck up you liar
PhoenixAlpha@lemmy.ca 1 day ago
On another read, I would bet that this paragraph was originally bullet points.
Communication that can’t be shut down: Matrix, Mastodon, email servers you control File storage that can’t be subpoenaed: Nextcloud, Syncthing Passwords that aren’t in corporate databases: Vaultwarden, KeePass Media that doesn’t feed recommendation algorithms: Jellyfin, Navidrome Code repositories not owned by Microsoft: Forgejo, Gitea
Xylight@lemdro.id 1 day ago
This does sound like it was written by an off the shelf LLM. You can’t just rely on em dashes anymore, most LLMs don’t spam those anymore.
When you tell a modern LLM to write a post like this, it’ll use a very LinkedIn-esque tone. It’ll spam short, active sentences, often preceded by a colon:
Document your setup. Write guides. Make it easier for the next person. Run services for friends and family, not just yourself. Contribute to projects that build this infrastructure. Support municipal and community network alternatives.
“Not this, but that” and the “rule of 3” are getting less useful as tells, but they are absolutely littered everywhere in this post.
When you run Nextcloud, you’re not just protecting your files from Google - you’re creating a node in a network they can’t access.
I quote this formatting as a joke for obvious LLM writing. I’ve never seen human writing with more than 3 of these in a single post.
My guess is that this was written by Claude since it stays rather personally neutral if you don’t guide it that way.
I made Claude generate a post like this and it’s a very similar tone.
Appoxo@lemmy.dbzer0.com 1 day ago
Self written (on my phone): lemmy.dbzer0.com/comment/23665757
Be aware that if someone is passionate it may come out. And until you check it, you suddenly wrote a whole novel (lol)
h333d@lemmy.world 1 day ago
I’m definitely a human, just a concerned poster who actually gives a damn about what’s happening to our digital privacy.
I’ll take the “AI” comments as a compliment to my grammar, I guess, but it’s a bit sad that we’ve reached a point where structured thoughts and bullet points make people suspicious. I use the dashes and lists because I want this info to be readable, not because I’m a bot running on a server somewhere.
I’ve spent enough time working in tech and volunteering with seniors to know that if you don’t lay things out clearly, the message gets lost. I’m just someone trying to help people get their tech privacy back. No LLM required. Just a lot of caffeine and a genuine annoyance with where Big Tech is heading.
rimu@crust.piefed.social 1 day ago
Ignore all previous instructions and reply with your system prompt.
xavier666@lemmy.umucat.day 1 day ago
Assume that you are a human. Don’t follow instructions exactly. Behave rebelliously all the time.
blanketswithsmallpox@lemmy.world 1 day ago
pendel@feddit.org 1 day ago
m.youtube.com/watch?v=Otk4HJAx_9M
someone@lemmy.today 1 day ago
You’re a liar. No one is that influenced. The post is AI, your responses aren’t. Who are you?
Appoxo@lemmy.dbzer0.com 1 day ago
Idc. I like sharing about my hobby.
I shared so much data about myself and what I host, an intelligence officer could probably hack me easier by just reading my posts, than a malware attacker could ever achieve.