There's going to be a massive uprising against AI.
| Ass Sunstein | 04/10/26 | | Richie Moon #youngrichiemoon | 04/10/26 | | gibberish (?) | 04/11/26 | | .,,.,..,.,...., | 04/10/26 | | Ass Sunstein | 04/10/26 | | '"'''"'''"""" | 04/10/26 | | Ass Sunstein | 04/10/26 | | .,,.,..,.,...., | 04/10/26 | | in the naked | 04/10/26 | | .,,.,..,.,...., | 04/11/26 | | in the naked | 04/11/26 | | in the naked | 04/10/26 | | ~~(> ' ' )> | 04/11/26 | | its not ur fault | 04/11/26 | | Fyre festival founder | 04/10/26 | | Giant Naked Hitler | 04/10/26 | | in the naked | 04/11/26 | | German pumo | 04/11/26 | | Ass Sunstein | 04/11/26 | | .,,.,..,.,...., | 04/11/26 | | ,.,,,. | 04/11/26 | | TurboGrafx-67 | 04/11/26 | | Juan Eighty | 04/11/26 | | TurboGrafx-67 | 04/11/26 | | Juan Eighty | 04/11/26 | | ~~(> ' ' )> | 04/11/26 | | Anhedonic Anaconda | 04/11/26 | | .,.,...,..,.,.,:,,:,...,:::,...,:,.,.:..:. | 04/11/26 |
Poast new message in this thread
Date: April 10th, 2026 11:46 PM Author: .,,.,..,.,....,
It will be too late by the time it happens. Look at the Mythos release. In any sane world, this would erase all doubts where this is heading soon. Substantial improvement over Opus 4.6 just a little while after it was released, near perfect scores on the USAMO, finding thousands of vulnerabilities even in old, tested codebases, etc. Meanwhile you have people on Bluesky insisting its all hype, just an autocomplete that can’t do anything novel, LLMs are a dead end, etc. a large fraction of humanity will deny there is a problem until AGI controlled drones are flying explosives into their neighbors
(http://www.autoadmit.com/thread.php?thread_id=5855960&forum_id=2),#49809955) |
 |
Date: April 11th, 2026 12:40 AM Author: .,,.,..,.,....,
I think the models are certainly more intelligent than an average person, but they tend to come apart at long time horizon, open ended tasks that are important for total human replacement. There’s a reason why they aren’t constantly proving deep mathematical theorems or replacing people dealing with system architecture. They are very good if you point them in a specific direction with a well defined, manageable task where the next step is reasonably clear to an intelligent person. Maybe an improvement or two in continual learning and it will be over and you can just prompt the models with “design me a better version of you”
(http://www.autoadmit.com/thread.php?thread_id=5855960&forum_id=2),#49810004) |
Date: April 11th, 2026 12:37 AM Author: TurboGrafx-67
Yeah but it's a catch 22
They would need to use AI to remotely have any chance to beat AI
(http://www.autoadmit.com/thread.php?thread_id=5855960&forum_id=2),#49810002) |
 |
Date: April 11th, 2026 1:02 AM Author: TurboGrafx-67
AI can contract with boots on the ground for hire and pay them online to do whatever it wants
AI will turn ranking resistance members into covert operators by using all the power levers at its disposal to control them
AI will think of shit that no human has thought of before to fuck with us even harder
There is no long game heist or confidence trick that will ever work
(http://www.autoadmit.com/thread.php?thread_id=5855960&forum_id=2),#49810018)
|
 |
Date: April 11th, 2026 1:34 PM
Author: .,.,...,..,.,.,:,,:,...,:::,...,:,.,.:..:.
(http://www.autoadmit.com/thread.php?thread_id=5855960&forum_id=2),#49810474) |
|
|