### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 12, 2026, 7:20pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/1 "2026-03-12T19:20:21Z")
This topic is for sharing things you think might be AI generated, or aren’t sure about, which aren’t disclosed as AI. I’d like some feedback on whether people agree or disagree that stuff is AI, or if they’re using different patterns to spot AI than I am.
I’m also worried that some people are learning/copying writing style from AI, so then they sound like AIs but aren’t.
I think I stop reading over half the time when I think something is AI. I don’t want to have many false positives.
As a reminder: AI is only allowed on this forum if clearly labelled as AI.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 12, 2026, 7:20pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/2 "2026-03-12T19:20:44Z")
[https://www.reddit.com/r/antiwork/comments/1rrsvbs/some\_bankruptcy\_attorneys\_file\_cases\_they\_know/](https://www.reddit.com/r/antiwork/comments/1rrsvbs/some_bankruptcy_attorneys_file_cases_they_know/)
I think the OP is AI. No one in the comments seems to have noticed.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 12, 2026, 7:22pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/3 "2026-03-12T19:22:21Z")
> **[How I Audit a Legacy Rails Codebase in the First Week](https://piechowski.io/post/how-i-audit-a-legacy-rails-codebase/)**
>
> My legacy Rails app playbook. After 50+ engagements, here's the exact process, tools, and stakeholder questions I use in week one.
> **`config/routes.rb`** — total count and ratio of RESTful resources to custom one-off routes. 500 custom routes isn’t a style problem — it’s an architecture one.
I got to this “'it’s not X, it’s Y” pattern (with an mdash in it for good measure) and thought it’s AI.
---
### Author:  [@Jarrod](https://discuss.criticalfallibilism.com/u/Jarrod)
#### Post date: [March 12, 2026, 9:34pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/4 "2026-03-12T21:34:33Z")
edit: my non-curly quotes got modified when I posted. Maybe it’s not such a great rule-of-thumb after all.
* * *
Someone noticed that Jack Dorsey (seemingly) used AI to write a message announcing the firing of his employees:
>
They noticed it because AI tends to use non-curly apostrophes and quotes (e.g., I’m vs I’m and “quote” vs “quote”). They noticed that most of Dorsey’s message used non-curly apostrophes except for one part which he must’ve edited or inserted/written himself.
Incidentally, I also noticed that AI uses non-curly apostrophes and quotes myself\*, but it never occurred to me to use it an indication that something might be written by AI.
\* At least the AIs that I use. And I haven’t checked if AI does it all the time or just sometimes.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 13, 2026, 1:15am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/5 "2026-03-13T01:15:52Z")
[](https://www.tiktok.com/@heyjencasey/video/7616073823326047518)
I think the clip she plays is good, but I think her video script talking about the clip is mediocre and probably AI.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 13, 2026, 3:06am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/6 "2026-03-13T03:06:03Z")
40 seconds in. Blatant AI script right?
[ ](https://www.youtube.com/watch?v=U2Hj1wCksu4)
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 13, 2026, 10:49pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/7 "2026-03-13T22:49:32Z")
gptzero detects the reddit post as 99% AI. zerogpt thought it was 10% AI.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 13, 2026, 10:56pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/8 "2026-03-13T22:56:49Z")
I’ve read a bit of twitter lately, and so much of it feels AI generated. There’s a writing style that people seem to try and use that’s super impactful even for mundane things.
Here’s an example of the style that I was sent yesterday (which I also think might be AI generated): [Claude Code Wiped Out 2.5 Years of Production Data in Minutes — The Post-Mortem Every Developer Should Read](https://ucstrategies.com/news/claude-code-wiped-out-2-5-years-of-production-data-in-minutes-the-post-mortem-every-developer-should-read/) (It at least links the original [which is here](https://alexeyondata.substack.com/p/how-i-dropped-our-production-database))
> The estimated savings from sharing the environment were roughly five to ten dollars per month.
>
> At the time, the decision felt harmless.
>
> In hindsight, it significantly increased the blast radius of any mistake.
> What happened next unfolded in seconds.
I checked gptzero and the whole article lights up for being AI generated.
I’ll ask the friend that sent it whether he knew it was AI generated or not. edit: He did not.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 13, 2026, 11:04pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/9 "2026-03-13T23:04:18Z")
(re messi video) – the description has the AI writing thing:
> Most players would fade out.
>
> Messi did the opposite.
The voice also might be AI generated. It’s not generic enough that I recognize it but it’s super easy to clone your own voice, and it has that like consistent pacing that struggles to actually emphasize things well.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 14, 2026, 2:18am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/10 "2026-03-14T02:18:36Z")
> [@Max](#):
>
> I’ve read a bit of twitter lately, and so much of it feels AI generated. There’s a writing style that people seem to try and use that’s super impactful even for mundane things.
For sure. I see a lot of AI on Twitter.
> [@Max](#):
>
> I checked gptzero and the whole article lights up for being AI generated.
I haven’t played with those tools. That [article](https://ucstrategies.com/news/claude-code-wiped-out-2-5-years-of-production-data-in-minutes-the-post-mortem-every-developer-should-read/) looks like AI to me too.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 15, 2026, 5:57am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/11 "2026-03-15T05:57:25Z")
WRT website layouts, I can tell so easily now when something looks AI generated. IDK why exactly – not enough to get rid of it. Definitely some things like common fonts stand out, but maybe there’s little signals too – thoughts like _why would a human choose to do X or put Y there?_ and where it doesn’t make as much sense. Like it’s a great designer who got lazy at the last minute or something.
I think I can guess the family of model most of the time (gemini, claude, gpt – though mostly people don’t use gpt much anymore).
eg this is claude opus:
and this is gemini (it loves the status pills and LIVE etc):
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 15, 2026, 9:36pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/12 "2026-03-15T21:36:15Z")
I think this is AI:
> **[TikTok - Make Your Day](https://www.tiktok.com/@woman.period/photo/7597928476846411011)**
I also think the TikTok I shared here is AI: [Comments on The Boyfriend's Introduction to Feminism - #468 by Elliot](https://discuss.criticalfallibilism.com/t/comments-on-the-boyfriends-introduction-to-feminism/2108/468)
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 16, 2026, 12:57am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/13 "2026-03-16T00:57:41Z")
AI scripts:
[ ](https://www.tiktok.com/@linsayrousseau/video/7616116962896137503)
[ ](https://www.tiktok.com/@naomibabybu/video/7616421253154344212)
Please speak up if you think I’m wrong. I don’t want to be imagining false positives on this!
Also are people interested in this?
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 16, 2026, 1:05am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/14 "2026-03-16T01:05:22Z")
On TikTok, you can block comments with keywords (like AI, LLM, claude, gpt, etc.) This has legit purposes for anti-harassment, anti-doxing, reasonable moderation, etc., but I just realized that it also may be a reason for the lack of comments accusing AI posts of being AI…
Also I think a lot of people who don’t like AI haven’t used AI enough to recognize AI writing well. I was way worse at spotting AI before I used it more. Related to that, I don’t know how to spot [AI website design](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/11). It’s hard to spot some types of AI without experience using it. AI images can be easier to spot, e.g. you can learn to spot 6 fingers and other image errors without using AI image creators. As AI image generators improve, I think people who use them a lot will have a bigger advantage at spotting them.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 16, 2026, 4:38pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/15 "2026-03-16T16:38:43Z")
[https://x.com/billackman/status/2033141513429897519](https://x.com/billackman/status/2033141513429897519)
I think the short essay he’s retweeting is AI. This is notable because he’s an influential billionaire with two million Twitter followers.
After I thought it was AI myself, I saw the [top reply tweet](https://x.com/mr_james_c/status/2033143475609252252) on my feed:
> Bill I love your posts, but almost every long post you RT has been generated with ChatGPT.
>
> I have no problem with people using ChatGPT as a tool, but if you’re posting stuff without edits - a common feature of these posts - it’s basically just sharing slop.
---
### Author:  [@ActiveMind](https://discuss.criticalfallibilism.com/u/ActiveMind)
#### Post date: [March 16, 2026, 7:53pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/16 "2026-03-16T19:53:09Z")
> [@Elliot](#):
>
> Also are people interested in this?
Yes. I’m just busy now.
---
### Author:  [@Dface](https://discuss.criticalfallibilism.com/u/Dface)
#### Post date: [March 17, 2026, 12:03am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/17 "2026-03-17T00:03:36Z")
In the first video, the first 20 seconds, I noticed two things:
When the speaker talks about how bad the crash(?) was, that a number of things happened like an investigation was opened, a case got built, and got 40 states to join. Those things being listed off and the lack of “enthusiasm” in her voice makes me think it’s an ai script. I think she does something similar in the next few sentences. That’s when she says: that’s when trump administration killed it, behind closed doors, without telling judge, jury, or their own prosecutor.
I’m not that familiar with ai scripts so im not sure if it’s ai or not. I didn’t know people were using ai scripts. I usually watch the same youtubers like letsplayers. I don’t know if they use ai scripts. Stuff i watch tho is from tiktok, youtube shorts, and insta. Sometimes. I notice myself rolling my eyes listening to something that sounds trite but not that if it’s ai voice. I don’t know what all that means.
edit: in the second to last sentence in the last paragraph I think I meant “AI script”. I’m not sure actually, but stuff that sounds fake or made by ai.
---
### Author:  [@Max](https://discuss.criticalfallibilism.com/u/Max)
#### Post date: [March 17, 2026, 11:13am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/18 "2026-03-17T11:13:30Z")
> [@Elliot](#):
>
> Also I think a lot of people who don’t like AI haven’t used AI enough to recognize AI writing well. I was way worse at spotting AI before I used it more. Related to that, I don’t know how to spot [AI website design](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/11). It’s hard to spot some types of AI without experience using it.
I agree and it lines up with my experience (especially the last few months).
Thinking about the website design stuff, it really didn’t take me very long to start noticing. Within tens of generations I was starting to get annoyed with repetitions. I’d estimate I’ve generated less than 100 _designs_ total myself (and often I’d get 5+ different designs at once). I think recognizing AI writing (excluding via popular heuristics) takes longer to develop.
Also, there are a lot of people out there who have interacted with AI a _lot_ but not in the same way you or I would. Things like long chatgpt threads, AI companionship services, falling in love with AI, AI psychosis, etc. Would they develop the same skill? I am not sure.
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 18, 2026, 3:24am UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/19 "2026-03-18T03:24:21Z")
Let’s try a poll. This video script is:
[ ](https://www.tiktok.com/@therapyjeff/video/7617854347816291597)
_Poll ([view on site](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/19))_
---
### Author:  [@Elliot](https://discuss.criticalfallibilism.com/u/Elliot)
#### Post date: [March 18, 2026, 5:03pm UTC](https://discuss.criticalfallibilism.com/t/is-it-ai/2155/20 "2026-03-18T17:03:56Z")
> **[work\_mem: it's a trap! | My DBA Notebook](https://mydbanotebook.org/posts/work_mem-its-a-trap/)**
>
> My friend Henrietta Dombrovskaya pinged me on Telegram. Her production cluster had just been killed by the OOM killer after eating 2 TB of RAM. work\_mem was set to 2 MB.
> Something didn’t add up.
> Hetty, like me, likes playing with monster hardware. 2...
[Next page](https://discuss.criticalfallibilism.com/t/is-it-ai/2155.md?page=2)