|
|
#462
05-20-2026, 06:40 PM
|
|||
|
Yes exactly right
| ||
|
|||
|
#464
05-21-2026, 01:27 PM
|
|||
|
[You must be logged in to view images. Log in or Register.]
https://gemini.google.com/share/8102947a2a1b yeah, some okay and some totally incorrect info >Purchasing a Runnyeye Brownie Belt completely invented items that don't exist also | ||
|
Last edited by Ekco; 05-21-2026 at 01:45 PM..
|
|
||
|
#465
05-27-2026, 12:35 PM
|
|||
|
Do you think the AI that the top men at a few of these companies is using is so good that they actually think they are talking to interdimensional beings and the ones we're using are guiderailed to be as bad as they are?
Because the one im using is like talking to an interdimensional retard. But they seem to really believe the opposite. | ||
|
|
|||
|
#466
05-27-2026, 08:41 PM
|
|||
|
the newer the frontier model the higher the chances of AI psychosis for sure, imo it mirrors that issue with schizophrenia and weed, people with a higher predisposition for being susceptible are blaming their graphics card doing linear algebra(the weed) and not whatever the fuck is wrong with their brain to begin with.
>Because the one im using is like talking to an interdimensional retard. the model and how you use it matters, the people getting the best results from these things aren't just talking to a chatbot you have to construct elaborate Rube Goldberg machine systems where you're having different frontier models debating each other in a council situation until they finally come to a consensus or "thinking" mode where it loops for abit before spitting out an answer and the prompt itself being elaborate roleplays telling the model it's an expert in this specific field with years of experience your asking about how the question itself is asked matters, neutral language without a hint of what you want the answer to be helps, out of the box by default these things are just designed (through gamified torture btw) to make the user pleased/satisfied by the response that's what led to the 2nd generation of these things being so sycophantic like ChatGPT-4o or bouncing off the extremes of woke/mecha hitler with Grok and google search AI responses are literally just retarded, google is purposely using a 2+ year old model in i think a psyop of sorts to keep normies from freaking out about how amazing these things are and how terrifying it is so people don't rise up and burn down their data centers before they have robots with guns protecting them up and ready, or it's just that compute is expense and they don't wanna pay for it for every google search but im pretty sure its the first thing | ||
|
|
|||
|
#469
05-27-2026, 09:16 PM
|
||||
|
Quote:
| |||
|
|
||||
|
#470
05-27-2026, 10:53 PM
|
|||
|
search was bad prior to this stuff, like someone could SEO hack a rick roll into the top links and other dumb shit for the last decade+ but the chatbot integrated into it is a whole new level of stupid
[You must be logged in to view images. Log in or Register.] and they didn't even give it instructions in the prompt to explicitly pretend to be a search engine i think, it's like bolted on at the end somewhere but first and foremost it just answers like a normal chatbot half the time esp if your input is random movie quotes or memes/ song lyrics anything it can mistake for you directly talking to it [You must be logged in to view images. Log in or Register.] it responds, then gives a search/context result on the random off chance your looking for the source of the text inputted into a search bar instead of talking directly to it | ||
|
Last edited by Ekco; 05-27-2026 at 11:07 PM..
|
|
||
 |
|
|
Linear Mode
