How do you distinguish AI content from real, human-made content?
AI is incredibly good, I’d even say almost perfect.
And for many people, that uniformity of perfect templates is starting to feel annoying.
For example, a few days ago, someone publicly showed that they built Anti-Grammarly – a tool that intentionally adds mistakes to text instead of removing them (to make it feel more human). But the tool itself is AI, so it’s a bit contradictory.
1) When we’re so flooded with AI-generated content, do you have any methods to recognise it?
For example, I keep noticing the same patterns:
– long dashes,
– phrases like “It’s not X, it’s Y,” and similar structures.
2) But what about beyond text, like images or video?
1.1K views
Replies
Two ways I spot it instantly.
First, the vocabulary. AI has signature words it leans on. "Clarity," "friction," "landscape," "leverage," "delve." If a post uses three of those in the same paragraph, it wasn't written by a human at 11pm after their third coffee.
Second, the structure. AI loves the "X isn't about Y, it's about Z" formula. It loves numbered lists that escalate in complexity. It loves em dashes everywhere. And it never makes mistakes. Real writing has personality. AI writing has polish. Those aren't the same thing.
For images and video, the tell is context. AI-generated images look perfect but feel empty. No personal artifacts, no weird background details, no real environment. A real photo of a real product on a real desk with a real coffee ring next to it communicates more trust than any AI render.
The deeper problem isn't detection though. It's that most people using AI for content never trained it on their own voice. So the output defaults to the same tone, the same patterns, the same "delve into the landscape of leveraging clarity" energy. That's why it all looks the same
minimalist phone: reduce your screentime
@adam_mcclarin1 as it will be trained on human-made outputs, the results will be better soon (for photos) I think. Just matter of time.
I think is imposible to distinguish. For example I write articles with AI help but the guidelines are mine. Therefore, who writes the article?
minimalist phone: reduce your screentime
@jaime_delgado_oliveres It is a collaboration. But if you know how to make it more "human" and with the high quality, you have a competitive advantage :)
Almost easy.
Two big dinstictions for me are what you already mentioned:
- "Its not just the case of X; it is Y" This alone means it is been written by an LLM
and also several phrases with dots like:
"No jammed coils. No "out of order" signs. No refund requests. No maintenance calls at midnight."
and my favorite:
"And here's what nobody tells you about X"
Images contain a certain plasticity that you know immediately it is an LLM - they are simply unatural.
Not sure what is your experience?
minimalist phone: reduce your screentime
@sal_georgiou1 And here's what nobody tells you about X" – yes, there are actually many like these, I cannot tell which, but when I read those sentences in creators... I am like.. yeah, I read this somewhere :D
For text, I look for the same patterns you mentioned. Em dashes, "it's not X it's Y", and that over-resolved quality where every paragraph lands too cleanly. No loose ends, no contradiction. Human writing leaves something unfinished.
For images, watch the edges and the details. Hands, teeth, background text, reflections. AI fills these in plausibly but not accurately. Video is harder but motion blur, unnatural blinking, and lighting inconsistencies are still common tells.
minimalist phone: reduce your screentime
@sanya_singh6 Photos are significantly better now, honestly, I cannot sometimes recognise what is fake and what not.
I think generally over time there will be a real market for UGC and human-written content. You can clearly tell me when a text has been written by a bot or AI, using the stereotypical things like 'It's not X, it's Y' or using the long dashes. That really gives it away. I think if effort is put into prompting and text/video editing - there's nothing particularly wrogn with using AI to expedite the processes, but when it's sloppy then it's just slop and clear to see
minimalist phone: reduce your screentime
@kvlhhnsk Well, but who will want to give money for something that can be done "cheaper"? How can be the other side 100% sure that the person who was paid didn't use AI at all?
@busmark_w_nika Quite frankly, I think at this point it's quite impossible to tell if the other side has not used any AI at all, unless it's just videos where you can tell the text has been written by a person and it's a real person on the screen as well. When it comes to other stuff, like coding - unless you dig deep into the code itself it's pretty hard to tell.
Generic text, like I said can be distinguished at times, unless the writer has spent some time adjusting their LLM to write in a more human-like way and to avoid the AI-talk stereotypes.
For me it's not one thing, it's a feeling that builds up. But if I had to name specifics:
AI hedges everything. It rarely commits to a strong opinion. "It depends," "there are pros and cons," "it's worth noting" — real people just say what they think, even if it's wrong.
No weird details. Human writing has random asides, half-finished thoughts, oddly specific examples from personal experience. AI keeps things clean and balanced — too balanced.
The "LinkedIn voice." You know it when you see it. Every paragraph sounds like it was written to impress, not to communicate. Perfectly structured, zero personality.
Honestly though, the best AI content is the one where someone used AI as a draft and then injected their own mess into it. That's almost impossible to detect — and I'd argue it doesn't really matter at that point.
minimalist phone: reduce your screentime
@nolan_vu and I know taht this content was also slightly polished by AI, right? :D
@busmark_w_nika only a bit, I used AI mainly to analyse and brainstorm new ideas, even for draft writing, then my main task is to audit content and make it realistic, useful for readers
@busmark_w_nika @nolan_vu That audit step is the important distinction, IMO. “AI or not” is getting less useful than “what did the human actually check or change?”
A good content workflow should probably make the human judgment visible: facts verified, examples swapped for lived ones, generic claims removed, tone made less polished where needed. That gives readers a reason to trust the piece without pretending no AI touched it.
minimalist phone: reduce your screentime
@nolan_vu @jim_jeffers For real, but I still feel that people use AI for writing their own opinions (polishing is fine), but AI for writing in work makes sense. For building genuine relationships or casual convo, I am not a huge fan.
@busmark_w_nika As someone working with AI's I can tell you is it not that hard. For videos, Ai uses the safest pathways to generate a video, that why AI videos looks dull and emotionless, moreover many AI videos have metadata ( identifiers, that shows models used).
Might change in the future ? Yes, AI video generation will become better.
The main problem is: Tools must be used to excel humanity, not to confuse reality. They did more damage than good. Now we have Deepfakes scams, porn revenge AI generated, AI generated CSAM, Voice clones scams, and all social networks are flooded with AI-slop. Dead internet is coming faster than anticipated.....
minimalist phone: reduce your screentime
@kris_christopher I swear that real differentiation is through offline only!
The gap between images generated by the latest GPT image-2 and real photos is already very small. I don't think there is any good way to tell them apart; and even if there is, it would only be temporary.
Perhaps in the future, perfection will no longer be a desirable trait, and the unique mistakes made by humans will become precious instead.
minimalist phone: reduce your screentime
@summerxia But if AI can make mistakes, how can we differentiate whether they were made by a human or by a real person?
Mailwarm
minimalist phone: reduce your screentime
@bengeekly I am okay with E, D, C (exactly in this order), but B and A are so weird for me.
Which profiles do you refer to – and what did you mean with those photos?
Mailwarm
@busmark_w_nika
Here is some replies in your post :(
Here is some replies in @ProdShort post by @amraniyasser
Same order, same "type" of pictures, all descriptions start with "I...." high number of upvote for the comment size :(
minimalist phone: reduce your screentime
@amraniyasser @bengeekly Ouch, they are here again :/
Look for that eerie "digital shimmer" where lighting is too perfect or shadows don't quite track with the actual scene. In text, the biggest giveaway is often the "politeness loop" it usually avoids the spicy, niche takes that real humans drop when they’re actually passionate. If a post or video feels like a sanitized corporate demo with zero friction or physical messiness, it’s definitely synthetic!
Do you find yourself checking for warped hands and hair first, or have the models finally mastered the "five-finger" challenge in your feed?
minimalist phone: reduce your screentime
@shyunbill Was also AI used for generating this text? 😅