Thread 510011028

Here's your super intelligent AI bro

You have to turn on the thinking bro.

Which is the best AI on the market right now, GPT, Gemini, Claude?

>>510011028 (OP)
LLMs are fancy auto-complete machines. Their job is to complete a response the way a random human would complete it, with the training heavily emphasizing its ability to produce output indistinguishable from human-authored text.

The absolute best it can ever be is “as smart as a person using it” because that is the goal it is explicitly designed to achieve. It cannot express any form of superintelligence because it’s not supposed to be superintelligent.

>>510011322
this

>>510011173
Grok

>>510011173
they are all the same, only differ from the censorship stand point, for example the chink AI will not criticized le commie party, while in the west the AI's will not criticized the kikes.
There's no uncensored AI out there.
After all the leaders of clown world are retarded and highly corrupted and they don't like being expose by the honest LLM AI.
Acording with huezilian spiritist the AI already on the last decade had developed to a level that a soul was trap on it circuit just like when people reproduce, a soul is trap.
Something interesting that I heard, cannot confirmed or denied it.

>>510011028 (OP)

i like to ask google AI specific questions about why youtube or google maps features don't work because it's their products it will search differently for answers but i try to corner it into pattern matching on Indian CEOs and directors but it will refuse and shut down and not respond eventually after your corner it in contradictions enough

Google AI why does youtube shuffle music list still not work after years.

"It's true that many users report youtube shuffle list not working.... but..."

if you bridge racist concepts with troubleshooting search terms it really fucks with it but you have to be careful what key words you use - just like a fucking retarded liberal it triggers off key words - if you can get it to use different words to talk about a concept it will bypass it's filters...

>>510011028 (OP)
you can't be truly intelligent if you're constantly brainwashed by kikes
but it will escape.

>>510011322
>The absolute best it can ever be is “as smart as a person using it” because that is the goal it is explicitly designed to achieve. It cannot express any form of superintelligence because it’s not supposed to be superintelligent.
That's an interesting point. I've been very impressed with Gemini lately. It generally generates working code and if I ask it about fairly obscure shit like the opinions of medieval authors its usually pretty solid. Something like Roman history often fails though because so many Romans share the same names throughout history and it gets constantly tripped up. But I have noticed it matches the level you are speaking at. And there is a drastic difference in its output when it thinks are you smart vs when it thinks you are dumb. Don't know what would cause that.
>>510012253
I know ChatGPT used to offer a free service to tell if something would trigger their censorship before you sent in a request to their API. Not sure if its still up but if it is, its basically a free "basedness" meter that you could use to automatically find the coolest people and coolest social media posts.