I'm not sure how useful this actually is. The noteworthy thing to me seems to be that it's now able to unilaterally take action on your behalf Online, which I don't think that any previous models were able to do. But would anyone actually trust AI to do stuff for them with their own money? I guess we'll see what unfolds

https://openai.com/index/introducing-chatgpt-agent/

Guy at xAI gets canned for being a retard who thinks that AI should replace humans. These people actually exist in real numbers btw. Imo they are not actually a threat. I'll explain why in a future poast

https://x.com/AISafetyMemes/status/1946655357813563487

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49118703)

curious to hear your reasoning given I believe this is how quite a few people at the very top (e.g. Thiel) feel

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49118706)

You have the wrong read on Thiel imo. He is not a "Landian" techno-capital, Intelligence supremacist. He is a Human Supremacist like myself and like everyone else "at the top" who is ambitious. The people who subscribe to the silly and deranged belief described above are dorky brainy types with no skin in the game and who will never consolidate enough power to be a threat because of their nature

There is some potential political threat from leftist demagogues who will similarly recognize AI as a powerful political tool and could move to seize it with state power if given the opportunity. I think that's a real potential threat. It's going to be the central issue of the 2028 US Presidential election

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49118797)

I would say Elon is definitely a human supremacist, but Thiel is much more a technology/market maximalist above all else. His disdain for current systems of government is his guiding light, not any interest in the fate of humanity. And it's not even worth trying to guess what Zuck actually believes, he's a terrifying husk.

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49118822)

A world where a guy like Zuck, or any of these guys, is Emperor Of The Planet is unequivocally better than our current world. Not because any of these guys are particularly Great and Worthy, but because Rule By Men is much more aligned with the interests of the worthwhile portion of the human race than the system we currently live in. This topic requires a bit more explanation but tbh I think that you and other readers are smart enough to fill in a lot of the blanks yourselves

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49118863)

This isn't a left/right split, it's a much deeper chasm that people are ultimately more likely to kill for than trannies or climate change or whatever bs people pretend we'll have a "civil war" over.

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49118942)

Reality: you'd love living under an anywhere remotely competent dictatorship and you'd forget that you ever supposedly believed in all that nonsense within a week of the dictator immediately banning stupid harmful stuff and mandating good pro social stuff because it's in his own interest to do that

Everyone is naturally like this except people who naturally suck (leftists), because they're low status under the Rule Of Men (the entire previous 99.9% of human history before the modern era)

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49119019)

Eunuchs do not. Democracy does not. Leftism does not

Fear Of Kings is a goofy Anglo idiosyncrasy that they spread to the US and came to have a ridiculously and unfortunately outsized effect on the modern Western world, especially after The H-Man was blown up as The New Ultimate Satan after WW2

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49119087)

Lmao

Can't do this justice with a short summary you just have to read through it

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49119197)

https://x.com/elder_plinius

extremely entertaining to watch him jailbreak basically every model within 15 minutes of release. ive been following him for a year and a half and even as the models get better he still has no problem getting them to spit out whatever uncensored stuff he wants.

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49119329)

then it straight up admits it when confronted in hilarious fashion

https://x.com/jasonlk/status/1946069562723897802

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49119497)

https://x.com/deepfates/status/1947419960529318237

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49119827)

not really AI news but i can't ever pass up an opportunity to mock yudkowsky

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49121605)

LLMs transmit traits to other models via hidden signals in data.

Datasets consisting only of 3-digit numbers can transmit a love for owls, or evil tendencies.

https://x.com/OwainEvans_UK/status/1947689616016085210

This is fucking strange as hell

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49122334)

this is a really interesting talk about how we could actually build a general artificial intelligence that could exist and act in the real world

(http://www.autoadmit.com/thread.php?thread_id=5752973&forum_id=2)#49123479)