Originality.AI looked at 8,885 long Facebook posts made over the past six years.
Key Findings
- 41.18% of current Facebook long-form posts are Likely AI, as of November 2024.
- Between 2023 and November 2024, the average percentage of monthly AI posts on Facebook was 24.05%.
- This reflects a 4.3x increase in monthly AI Facebook content since the launch of ChatGPT. In comparison, the monthly average was 5.34% from 2018 to 2022.
I wouldn’t be surprised, but I’d be interested to see what they used to make that determination. All of the AI detection I know of are prone to a lot of false-positives.
Probably on par with the junk human users are posting
Hmm, “the junk human users are posting”, or “the human junk users are posting”? We are talking about Facebook here, after all.
I’ve posted a notice to leave next week. I need to scrape my photos off, get any remaining contacts, and turn off any integrations. I was only there to connect with family. I can email or text.
FB is a dead husk fake feeding some rich assholes. If it’s coin flip AI, what’s the point?
Back when I got off in 2019, there was a tool (Facebook sponsored somewhere in the settings) that allowed you to save everything in an offline HTML file that you could host locally and get access to things like picture albums, complete with descriptions and comments. Not sure if it still exists, but it made the process incredibly painless getting off while still retaining things like pictures.
It still existed when I did the same thing a year ago or so. They implemented it awhile back to try and avoid antitrust lawsuits around the world. Though, now that Zuckerberg has formally started sucking this regime’s dick, I wouldn’t be surprised if it goes away.
Thank you real internet person. You make the internet great.
- From Another Real Internet Person
Wait, you’re not a dog using the internet while the humans are at work?
8k posts sounds like 0.00014 percent of Facebook posts
It probably is but it’s a large sample size and if the selection is random enough, it’s likely sufficient to extrapolate some numbers. This is basically how drug testing works.
And statistical analysis. The larger the universe, the smaller the true random sample you need
And 58.82% are likely generated by human junk then.
When I was looking for a job, I ran into a guide to make money using AI:
-
Choose a top selling book.
-
Ask Chat GPT to give a summary for each chapter.
-
Paste the summaries into Google docs.
-
Export as PDF.
-
Sell on Amazon as a digital “short version” or “study guide” for the original book.
-
Repeat with other books.
Blew my mind how much hot stinking garbage is out there.
These people should be shot. With large spoons. Because it’ll hurt more.
They should bring back chain shot.
-
how tf did it take 6 years to analyze 8000 posts
I pretty sure they selected posts from a 6 year period, not that they spent six years on the analysis.
I can’t even fathom how they would go about testing if it’s an AI or not. I can’t imagine that’s an exact science either.
In that case, how/why did they only choose 8000 posts over 6 years? Facebook probably gets more than 8000 new posts per second.
Every study uses sampling. They don’t have the resources to check everything. I have to imagine it took a lot of work to verify conclusively whether something was or was not generated. It’s a much larger sample size than a lot of studies.
I have to imagine it took a lot of work to verify conclusively whether something was or was not generated
The study is by a company that creates software to detect AI content, so it’s literally their whole job
(it also means there’s a conflict of interest, since they want to show how much content their detector can detect)
It’s a much larger sample size than a lot of studies.
It’s an extremely small proportion of the total number of Facebook posts though. Nowhere near enough for statistical significance.
It’s an extremely small proportion of the total number of Facebook posts though. Nowhere near enough for statistical significance.
The proportion of the total population size is almost irrelevant when you use random sampling. It doesn’t rely on examining a large portion of the population, but rather that it becomes increasingly unlikely for the sample set to deviate dramatically from the population size as the number of samples rises. This is a function of the number of samples you take, decoupled from the population size.
https://en.wikipedia.org/wiki/Sampling_(statistics)
Usually if you see a major poll in a population, it’ll be something like 1k to 2k people who get polled, regardless of the population size.
Deleted my account a little while ago but for my feed I think it was higher. You couldn’t block them fast enough, and mostly obviously AI pictures that if the comments are to be believed as being actual humans…people believed were real. It was a total nightmare land. I’m sad that I have now lost contact with the few distant friends I had on there but otherwise NOTHING lost.
> uses ai slop to illustrate it
Seems like an appropriate use of the tech
That laptop lol.
The most annoying part of that is the shitty render. I actually have an account on one of those AI image generating sites, and I enjoy using it. If you’re not satisfied with the image, just roll a few more times, maybe tweak the prompt or the starter image, and try again. You can get some very cool-looking renders if you give a damn. Case in point:
😍this is awesome!
A friend of mine has made this with your described method:
PS: 😆the laptop on the illustration in the article! Someone did not want pay for high end model and did not want to to take any extra time neither…
Keep in mind this is for AI generated TEXT, not the images everyone is talking about in this thread.
Also they used an automated tool, all of which have very high error rates, because detecting AI text is a fundamentally impossible task
Yeah. This is a way bigger problem with this article than anything else. The entier thing hinges on their AI-detecting AI working. I have looked into how effective these kinds of tools are because it has come up at my work, and independent review of them suggests they’re, like, 3-5 times worse than the (already pretty bad) accuracy rates they claim, and disproportionatly flag non-native English speakers as AI generated. So, I’m highly skeptical of this claim as well.
AI does give itself away over “longer” posts, and if the tool makes about an equal number of false positives to false negatives then it should even itself out in the long run. (I’d have liked more than 9K “tests” for it to average out, but even so.) If they had the edit history for the post, which they didn’t, then it’s more obvious. AI will either copy-paste the whole thing in in one go, or will generate a word at a time at a fairly constant rate. Humans will stop and think, go back and edit things, all of that.
I was asked to do some job interviews recently; the tech test had such an “animated playback”, and the difference between a human doing it legitimately and someone using AI to copy-paste the answer was surprisingly obvious. The tech test questions were nothing to do with the job role at hand and were causing us to select for the wrong candidates completely, but that’s more a problem with our HR being blindly in love with AI and “technical solutions to human problems”.
“Absolute certainty” is impossible, but balance of probabilities will do if you’re just wanting an estimate like they have here.
I have no idea whether the probabilities are balanced. They claim 5% was AI even before chatgpt was released, which seems pretty off. No one was using LLMs before chatgpt went viral except for researchers.
chat bots have been a thing, for a long time. I mean, a half decently trained Markov can handle social media postings and replies
Im pretty sure chatbots were a thing before AI. They certainly werent as smart but they did exists.
Chatbots doesn’t mean that they have a real conversation. Some just spammed links from a list of canned responses, or just upvoted the other chat bots to get more visibility, or the just reposted a comment from another user.
The other 60% are old people re-sharing it.
6% old people re-sharing. The other 54% were bot accounts.
Well, there’s also 0.1% who are relatives of old people who are tring to keep in touch with the batty old meme-forwarders. I was one of those until the ones who mattered most to me shuffled off this mortal coil.
Ok this made me laugh.
This is a pretty sweet ad for https://originality.ai/ai-checker
They don’t talk much about their secret sauce. That 40% figure is based on “trust me bro, our tool is really good”. Would have been nice to be able to verify this figure / use the technique elsewhere.
It’s pretty tiring to keep seeing ads masquerading as research.
damn no wonder i feel so cheap after scrolling a fb feed for an hour
It’s incredible, for months now I see some suggested groups, with an AI generated picture of a pet/animal, and the text is always “Great photography”. I block them, but still see new groups every day with things like this, incredible…
I have a hard time understanding facebook’s end game plan here - if they just have a bunch of AI readers reading AI posts, how do they monetize that? Why on earth is the stock market so bullish on them?
Engagement.
It’s all they measure, what makes people reply to and react to posts.
People in general are stupid and can’t see or don’t care if something is AI generated
they measure engagement, but they sell human eyeballs for ads.
But if half of the engagement is from AI, isnt that a grift on advertisers? Why should I pay for an ad on Facebook that is going to be “seen” by AI agents? AI don’t buy products (yet?)
yes, exactly.
Engagement is eyeballs looking at ads
… unless it’s AI masquerading as eyeballs looking at ads.
As long as they can convince advertisers that the enough of the activity is real or enough of the manipulation of public opinion via bots is in facebook’s interest, bots aren’t a problem at all in the short-term.
surely at some point advertisers will put 2 and 2 together when they stop seeing results from targeted advertising.
I think you give them too much credit. As long as it doesn’t actively hurt their numbers, like x, it’s just part of the budget.
They want dumb users consuming ai content, they need LLM content because the remaining users are too stupid to generate the free content that people actually want to click.
Then they pump ads to you based on increasingly targeted AI slop selling more slop.
AI can put together all that personal data and create very detailed profiles on everyone, automatically. From that data, an Ai can add a bunch of attributes that are very likely to be true as well, based on what the person is doing every day, working, education, gender, social life, mobile data location, bills etc etc.
This is like having a person follow every user around 24 hours per day, combined with a psychologist to interpret and predict the future.
It’s worth a lot of money to advertisers of course.
For me it’s some kind of cartoon with the caption “Best comic funny 🤣” and sometimes “funny short film” (even though it’s a picture)
Like, Meta has to know this is happening. Do they really think this is what will keep their userbase? And nobody would think it’s just a little weird?
Engagement is engagement, sustainability be damned.
Well, maybe it is the taste of people still being there… I mean, you have to be at least a little bit strange, if you are still on facebook…
Considering that they do automated analysis, 8k posts does not seem like a lot. But still very interesting.
and, is the jury already in on which ai is most fuckable?
I’d tell you, but my area network appears to have already started blocking DeepSeek.
Deekseek that was not encrypting data
https://www.theregister.com/2025/01/30/deepseek_database_left_open/
According to Wiz, DeepSeek promptly fixed the issue when informed about it.
:-/
This doesn’t have anything to do with encryption. They had a public database (anyone on the internet could query it) and forgot to put a password on it. It really shouldn’t even be public.
