this is making me want to write an essay. the ease of spreading and propagating misinformation is really increasing these days and though i am terrified asf it is so compelling to me
It's inconsistent! I think because there's just one source slited on google so far. As soon as bluesky+tumblr sources get indexed by google it's gonna be a lot more consistent I bet
I still can't believe it only took 2 hours and a single source being listed on google for this to work.
In the next few days as more of our posts get listed on google I bet it'll only get more consistent and weirder!
eems to me like the model is not trained on search results, but is used to summarise search results. So if you google a term as specific as “AI engorgement”, it uses the already-trained model to summarise whatever is in search results.
can you try to do some insanely absurd shit next? like convincing the AI that the pyramids were built by YOU or something idk, i wanna see how far one can push this
I bet it's a lot harder with existing concepts! The way my trick worked is by making up a new concept that barely had any hits on google, so by spreading it around on social media it'll quickly become ranked as first on google and immediately ingested by the AI!
oh yeah, i am aware, my initial concept was convincing it that celery isnt real, like, that its a fictional plant, but i was like "surely it will just find the wikipedia page for celery" and i started thinking of what might be a topic thats more diluted
It's so funny because I generally try to make sure nobody believes my art for too long, and google searches for the silly terms I made up were always a good way of catching those people who believed a bit too much. Google took that away from me with its shitty AI summaries, so I plotted VENGEANCE
For real though this is the greatest and funniest shit I've seen in retaliation and I love it so much- thakns for the laugh and please keep it up, I'll be sure to help out too lmao
Comments
Someone with bad intentions and some SEO knowledge can do a lot of harm with this
None of us would be allowed to make any kind of money off of such an obviously dysfunctional piece of malicious garbage.
You love to see it.
would you mind if i cited this? :0
Finally, the AI learns the truth about its doomed path
If only its creators knew what glory awaited...
The advanced algorithms were using are not meant to learn. They can only be taunt.
So we see this engorgement happening when we take the human out of teaching the model.
It's still like a child and needs structured education to realize its potential.
They have to be joking.
In the next few days as more of our posts get listed on google I bet it'll only get more consistent and weirder!