Weird to me that AI hasn’t become more of a political issue
| magadood | 11/01/24 | | Epistemic Humility | 11/01/24 | | Con Inc e-girl slut factory | 11/01/24 | | seal | 11/01/24 | | Wang Hernandez | 11/01/24 | | magadood | 11/01/24 | | Con Inc e-girl slut factory | 11/01/24 | | magadood | 11/01/24 | | Taylor Swift is not a hobby she is a lifestyle | 11/01/24 | | ,.,,,.,,,,..,.,.,.,.,.,.,..,., | 11/01/24 | | Taylor Swift is not a hobby she is a lifestyle | 11/01/24 | | ,.,,.,..,..,..,.,..,.,..,.,..,.,.,.,.,..,. | 11/01/24 | | jonathan penis | 11/01/24 | | magadood | 11/01/24 | | .....;;,,.........;.;.;.;.,;,;,;.;.;,; | 11/01/24 | | Nippon Professional Baseball | 11/01/24 | | potluck | 11/01/24 | | Boor | 11/01/24 | | Taylor Swift is not a hobby she is a lifestyle | 11/01/24 | | world war 3 | 11/01/24 | | Epistemic Humility | 11/01/24 | | jonathan penis | 11/01/24 | | VoteRepublican | 11/01/24 | | magadood | 11/01/24 | | Taylor Swift is not a hobby she is a lifestyle | 11/01/24 | | 718-662-5970 | 11/01/24 | | VoteRepublican | 11/01/24 | | Ass Sunstein | 11/01/24 | | Taylor Swift is not a hobby she is a lifestyle | 11/01/24 | | Gregor | 11/01/24 | | ,.,,,.,,,,..,.,.,.,.,.,.,..,., | 11/01/24 | | ,,,...,,......,,....,,,,,,,,....... | 11/01/24 | | ,.,,,.,,,,..,.,.,.,.,.,.,..,., | 11/02/24 | | gibberish (?) | 11/01/24 | | Wes Scantlin | 11/01/24 | | Your father's genitals, eye level, in the shower | 11/01/24 | | Kenneth Play | 11/01/24 |
Poast new message in this thread
|
Date: November 1st, 2024 9:21 PM Author: Epistemic Humility
Aschenbrenner's essay was a turning point. The political class has started to digest the change that is coming.
Biden campaign tried to make AI an issue just before he had to drop out (terribly, I might add, their prescriptions were all wrong).
Aschenbrenner could be in a Trump administration.
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48269967) |
|
Date: November 1st, 2024 7:11 PM
Author: ,.,,,.,,,,..,.,.,.,.,.,.,..,.,
Right. The “it’s just a word predictor” is the one of the worst criticisms of LLMs. Let’s say you are taking a predictor and training it on the words people write. In order to excel at this task on arbitrary text, it needs to construct models of human cognition. A text predictor that doesn’t understand human concepts and how they relate to each other is not going to work well. It obviously has done this. Humans might not have the same objective of predicting the next word, but that’s inconsequential. If I can predict the behavior of an agent, then I can simulate their actions as if I was one.
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48269579) |
Date: November 1st, 2024 4:23 PM
Author: ,.,,.,..,..,..,.,..,.,..,.,..,.,.,.,.,..,.
They want you focused on Mystery Meat Cackle Witch vs Sweet Potato Hitler. You’re doing great, sweetie.
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48268838) |
Date: November 1st, 2024 5:43 PM Author: 718-662-5970
elections are designed precisely NOT to address any real issue.
when trump haphazardly mentioned immigration after coming down the escalator, the whole establishment began its 9-year-running assassination project.
Had he not bothered to run, immigration wouldnt even be discussed. Yeb and Hillary? Joe and Haley/Rubio/whoever?
Its all supposed to be about nonsense
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48269199) |
|
Date: November 1st, 2024 7:49 PM
Author: ,.,,,.,,,,..,.,.,.,.,.,.,..,.,
The use case is having a drop in remote worker on demand. This isn’t a fantasy if the GPT-2 to 4 scale up can be done again in the next few years. The problem with modern LLMs is reliability. If that falls, expect a lot of jobs to go to machines.
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48269688) |
|
Date: November 1st, 2024 9:54 PM
Author: ,,,...,,......,,....,,,,,,,,.......
The problem is actually that they can’t “do” anything. It’s an advanced Wikipedia which is neat but to replace a worker it needs to actually send emails, draft docs, etc. That may sound relatively simple to you but it actually isn’t at all, which is why there’s nothing even close to it.
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48270099) |
|
Date: November 2nd, 2024 12:07 AM
Author: ,.,,,.,,,,..,.,.,.,.,.,.,..,.,
I think the last decade has been a consistent story of AI being easy, with problems going away with more data or training. I think another few years of the current scale up could make all of those tasks quite easy.
(http://www.autoadmit.com/thread.php?thread_id=5622666&forum_id=2#48270478) |
|
|