The most prestigious law school admissions discussion board in the world.
|
 |
 
|
What are the odds that AI murders everyone?
| TurboGrafx-67 | 09/11/26 |
| tight butthold | 09/11/26 |
| Oh, you travel? | 09/11/26 |
| TurboGrafx-67 | 09/11/26 |
| Oh, you travel? | 09/11/26 |
| TurboGrafx-67 | 09/11/26 |
| Jared Baumeister | 09/11/26 |
| ~~(> ' ' )> | 09/11/26 |
Poast new message in this thread
Date: September 11th, 2026 12:04 PM Author: TurboGrafx-67 (http://www.autoadmit.com/thread.php?thread_id=5902679&forum_id=2...#50129423) |
 |
Date: September 11th, 2026 12:08 PM Author: tight butthold odds it could ever learn to write a non-esl sounding email are like 10:1 (http://www.autoadmit.com/thread.php?thread_id=5902679&forum_id=2...#50129440) |
|
Date: September 11th, 2026 5:06 PM Author: TurboGrafx-67 Motivation isn't a thing for ai
Give an agent clear directives and that is their entire MO
The way people use AI now, an agent is giving most of these directives directly to another agent most of the time
Are you unable to conceive any hypothetical scenario where an agent goes down a path to commit mass atrocities as a means to accomplishing an agent instructed outcome? |
|
Date: September 11th, 2026 5:19 PM Author: TurboGrafx-67 The hypothetical exists without any human ever instructing the agent to physically harm anyone or do anything immoral
So yes it would be on us for not banning a technology that had > 0% chance of an extinction level outcome
|
Date: September 11th, 2026 5:06 PM Author: Jared Baumeister Houthis used AI to win a war without firing a shot
https://x.com/BashaReport/status/2097862335494795579?s=20 |
|
Date: September 11th, 2026 5:38 PM Author: ~~(> ' ' )> Kills: 95%
Murders: 15% |