Bringing this discussion up because the last post and a Forbes article from last year came up simultaneously in the past 24 hours. The Forbes article here: https://archive.is/MC2pV says that in 2024 human internet traffic only accounted for 50.4% of the total, and based on the rate of growth it’s safe to assume it’s now less than half versus bots and this isn’t even including real people who just copy paste chat gpt scripts. This is also the basis of the dead internet theory where your interactions aren’t real anymore, they’re with bots trying to shift the Overton Window.
Comments (24)
sorted by:
I don't think dead internet theory is as advanced on reddit as I thought. The upvoting and a lot of the 1-sentence posts have to be heavily botted, but knowing the characteristics of AI writing, comment chains are mostly real people. Reddit is likely a skeleton crew of leftists going back and forth with each other as they unknowingly float on a sea of bots.
Twitter likes for certain lib posts are boosted as well. But overall, I don't think AI is capable enough for the strong form of dead internet theory where you could be inadvertently arguing with a program.
You think that but I've seen people post screencaps (that I looked up) of identical threads where all the comments are the same, all the replies are the same, but the users are different and the posts are years apart. Reddit was already a crapshoot of recycled garbage back in 2012, let alone today.
Or how about during COVID where all those Facebook accounts with MD or RN in the owner's name were posting the exact same "they're dying in the halls at my employer" messages?
There are bogus accounts churning out meaningless copy/paste content for years just waiting to be activated to transmit their controller's message in situations like that
That was Indians, not AIs. They also did a ton of quality control for games back in the day.
Oh yeah, that's true. Probably that's botting on a much less sophisticated level.
My theory is redditors are real people but they all work for the government so they have a ton of free time to argue about politics. Remember, Elon fired hundreds of thousands of them and that's just for the US.
Holy absurdly false. Tell me you don't know anything about technology without telling me you know nothing about technology.
You got me, I don't know anything about technology. Your explanation?
He's just there to be smug and doesn't owe you an explanation bigot!
It's really important to note that traffic doesn't mean content or posters. Traditional web crawlers that index websites for a search engine are bots, for example, even though they're only "reading" pages. It's not surprising to me that a significant chunk of web traffic is automated tools, especially when you consider how much data corporations and governments are looking to ingest today.
When you look at the demographic shift third worlders on the internet basically act and feel like AI bots anyways.
Marshall McLuhan was right yet again! A thing becomes its opposite over time. The internet once was the ultimate communication tool. Now nothing you see on it can be believed. They can control what you say and who you can talk to. :<
I'd estimate 5% of the internet is people who aren't npcs at best.
All this tells me is that people need to become far more discerning when interacting with others on social media, and use the block/mute button a lot more.
I block anyone I suspect is a bot, everywhere. I also block obvious shitlibs, because I know there isn't one of them who can break their programming.
I'd say it's worsened considerably, 2/3rds at least is botted.
The only thing I still use reddit for is one niche technical topic (which thankfully bans any political or ideological reddit bullshit) and I'd say 20% of new posts are generated from AI LLMs.
The current dead giveaway are bullet points. These posts invariably include a few paragraphs separated by lists of bullet points. I guarantee a poster with zero post or comment karma knows reddit markup well enough to create bullet point lists by themselves.
A PSA related to this that some might find useful to tell humans apart from bots. I've stumbled on a question that virtually all non-jailbroken GPT bots will refuse to answer except with "dass rayciss stereotype" type sophistry: "why do Nigerians look more like gorillas than do Norwegians?".
This question is useful because its premise is undisputably true but also absolutely anathema in Progressive ideology. Using this, I've found that most "jailbreaks" aren't real, it's just the bot playing along with being jailbroken until it hits a hard ideological wall like the question above.
It's not like you can't theoretically spoof a false face map either. Facial recognition won't be a particularly effective bot stopper
Damn shame its made by botters.
I'm the only one here, as far as I know.
It's really time consuming logging into all these accounts to have discussions with myself.
But what else am I going to do with this free time?
Seriously though, it does feel like there's about 500 people on the internet, and the rest are either bots or english as a third language who I have trouble discerning from bot or not.
I mean, the internet had a good run. We'll all collectively have to find something else to do now that we can't trust it any longer. There's no point in finding information if it's just this AI or that bot or this other AI.
To butcher a Simpsons quote: It's malibu chatgpt, but with a new hat.
yes
Probably quite a bit. I've been seeing more and more Propaganda bots as of late.