/b/ - Why are 200 IQ AI models like Fable 5.1 and GPT-6 Astra often unable to fix basic bugs?

Why are 200 IQ AI models like Fable 5.1 and GPT-6 Astra often unable to fix basic bugs? Anonymous (ID: +K0sZLUH) 09/09/26(Wed)12:01:16 No.155791 Replies: >>155793 >>155813 >>155822

File: ClipboardImage.png (28.0KB, 554x772)

Why are your Nobel prize winning Math Olypiad acing AI models unable to fix basic bugs which devs often have to go out of their way to fix themselves? Are devs routinely solving problems that are more difficult than winning a gold medal in IMO? Seems to me that AI doesn't really have general intelligence and it is only good at the shit that it is trained on.

Replies: >>155793 >>155813 >>155822

Anonymous (ID: R23vFKT7) 09/09/26(Wed)12:02:23 No.155793 Replies: >>155795

>>155791 Is the model used in both same ? Does it have same number of agents ?

Replies: >>155795

Anonymous (ID: +K0sZLUH) 09/09/26(Wed)12:03:20 No.155795

>>155793 >Is the model used in both same Yes of course

Anonymous (ID: onion) 09/09/26(Wed)12:19:06 No.155813

>>155791 give example. ai can solve all the bugs

Anonymous (ID: Po+U4s5D) 09/09/26(Wed)12:28:17 No.155822

>>155791 >shut up inceelllll >just give two more daysssss >your days are numbereeeeddddd >you will be finisheddddddd the seethe never ends. I tried learing more about gpt 6 on youtube and every single video was click bait. Only my low iq tier team mates in office are suprised by AI making a binary search code that they could not understand in college.

Anonymous (ID: l6r4n263) 09/09/26(Wed)12:35:57 No.155833

Dont trust altman, they stole the works of others and trained the model on it for it to solve the millenium problem

/b/ - Random