It took a while for me to realize that pic related was actually two AIs talking to each other. Unfortunately, I feel that only reflect just how low a bar humanity sets. We need some way to test this more rigorously.
So, I propose the following: we make an AI to post on /pol/. Every couple of hours it makes a new post, except replies have a 50% chance of being responded to within minutes. Afterwards, we collect threads where the AI participated and ask people to guess which posts were made by actual people vs an AI.
We could try some variations on the above, such as making specific AI threads on /pol/ where the trick will be to distinguish real AIs from trolls, or even staging entire AI arguments on /pol/ threads and seeing how long it takes the /pol/lacks to figure out what's going on.
If nothing else, it'll be amusing to watch.
Are you inciting /sci/ to raid /pol/ ?
raids are against the rules
why do people think that letting two shitty AI's talk to each other is somehow going to make any kind of progress towards an actual real deal AI? you can't fabricate human-like consciousness using computer hardware in the same way you can't build a skyscraper with cake ingredients.
>It took a while for me to realize that pic related was actually two AIs talking to each other.
First day using the internet, huh, Grandpa. Turns out if you repeat a bunch of random text that someone sent to you, you look vaguely like a thing that isn't completely and utterly constructed from a pseudorandom number generator.
>>9086542
That's that stupid Googlebot it just parrots stuff. Hardly AI.
Also over half the stuff on /pol/ are blacked bots or antifa subversion threads.
It's the most spammed and attacked board on 4chan.
Your posts will be a drop of piss in an ocean of piss. Your likely replies will be from other trolls