sofiechan home

Does Leopold Aschenbrenner make any sense?

anon_zutu said in #1926 1y ago: received

Leopold Aschenbrenner has been everywhere for the last couple of weeks, feeding on the AI hype, doomerism and China hawkishness. He has very good credentials, but does any of his takes make any sense?

Link to his paper: https://situational-awareness.ai/

Leopold Aschenbrenne received

anon_pico said in #1927 1y ago: received

This image is nonsense. There is no map between GPT-X and the development of a human. The difference is in the size of post-compression data and techniques to better decompress data.

It makes sense that the US security apparatus wants to control novel technology and will utilize spurious reasoning to that end.

This image is nonsen received

anon_swra said in #1929 1y ago: received

>>1927
If one takes "smarter" or "human development", as he does, to mean the ability to answer standardized test questions which have known solutions, that is probably true and something that parents could take into consideration in their children's education. Especially in a place like California where quality does not rise in step with expense.

>spurious reasoning
If one really digs for the origin of the "han chinese" as a concept one ends up looking at the anti-foreign sentiment that swept across china in the late 19th century. Both the PRC and ROC regularly display images of Sun Yat-sen is the legacy of the May 4th movement, demonstrating the real taboo. But the real value of the China hawkishness is to use the pointing at the mulberry tree to scold the locust tree (about grievance ideology or other topics).

If one takes "smarte received

xenophon said in #1930 1y ago: received

>>1927

> This image is nonsense. There is no map between GPT-X and the development of a human.

Mira Murati made very similar comments in a recent interview, so this seems to be a trope within OpenAI.

If you take the comparison to mean "able to perform many of the online tasks that you might give to such a person" and not "actually equivalent to the human," then the comparison might make some sense.

Mira Murati made ver received

anon_xogu said in #1931 1y ago: received

>>1926
I think he's right about security and wrong about most other things. It's true that silicon valley isn't taking info security seriously. Top AI executives have girlfriends from that one university in Beijing that they recruit spies from, no doubt many spies are all through these companies, and there are constant news stories about China exfilarating western technology. If you don't want to just be a dev lab for Chinese power, you have to think about this.

As for his AI hype, he's off the charts. Claiming that the difference between gpt-4 and an independently creative PhD level AI researcher is 6 orders of magnitude of compute, and that this will happen in the next 5 years seems naive to the point of stupidity. Even equivalences between current gpt and any level of human intelligence are totally spurious. A cat can do complex acts of planning and independent thinking that gpt can't.

He also claims that automated AI researchers would speed things up and we would get recursive self improvement, but I don't see how this squares with his emphasis on the primacy of effective computing power. Conceptual breakthroughs probably would get us closer to actual AGI much faster, but I'm not seeing his schema cutting reality at the joints there.

Finally we have the arms race claims, which are always obfuscating. The arms race is between western labs seeking funding and attention. The Chinese are just following that lead. What's the big Chinese AI lab that the west has to beat? No one ever answers this question.

He seems to be sent by Holden Karnofsky's camp to increase their power in the AI scene and invoke the security state to take an increasing role. That might or might not be a good idea; I don't know their plans.

I think he's right a received

anon_dava said in #1933 1y ago: received

No, he doesn't make any sense, which is why it's so so ridiculous that anyone takes him seriously. Especially because they have a precedent to go off on: the dude's own terrible COVID19 prediction, in which he believed the world was going to collapse because he didn't account for something as obvious as human reaction and adaptation. That's an aside.

The one big actual problem with his prediction:
— Unknown unknowns. Anyone who has ever done anything new IRL knows that some things cannot be accounted for because you don't even know that you don't know that they exist. You only realize why you couldn't have made the correct prediction after the fact.

I think one reason some people take him seriously is because of his credential, which in my book actually counts against him. To me, doing exceptionally well at school means a person shouldn't be taken that seriously at all. Because they were happy to satisfy the silly requirements of midwit professors. Basically, I think very lowly of contemporary academics in general. I do not believe any respectable person who believes in their own competence would ever subject themselves to the nonsense you need to go through to become an academic. Therefore, becoming an academic already is a big indictment of a person. Being successful at satisfying those kinds of people who become academics so that you graduate top of your class is a double indictment.

No, he doesn't make received

anon_ryda said in #1934 1y ago: received

>I think one reason some people take him seriously is because of his credential, which in my book actually counts against him

Yeah, he definitely is a striver type, but that means he's remarkably good at distilling the current thoughts of the rich and powerful of a particular industry and playing to their emotions in a way that enhances his own standing among them or encourages them to invest in him. There's definitely some utility in reading (summaries) of his writing because of this.

Does he justify his actions based on effective altruism? If so, he's probably been co-opted into the ideological preferences of some variant of utilitarianism and it's definitely justified to obtain such resources in order to redistribute them altruistically. Of course, the problem always being what exactly you use to determine the sum total of well-being or what makes people happy.

Yeah, he definitely received

anon_hoku said in #1935 1y ago: received

I looked at the post a while ago, and stopped reading pretty quickly. Seemed heavy on hypnotic language and light on argument, and I didn't expect to learn anything.

>There's definitely some utility in reading (summaries) of his writing
Anyone wanna summarize it, then? Or if I just go with the apparent consensus here, "it's a grift playing off the fears of spies and financiers", will that give me everything important?

I looked at the post received

anon_kyji said in #1977 14mo ago: received

Leo is clearly a fag bottom, he sold his hole to Sam Altman in exchange for a fake credential. Some Indian putting him on his podcast is nothing. Can we forget this all after pride month?

Leo is clearly a fag received

anon_xogu said in #1986 14mo ago: received

>>1935
The summary is basically "AGI soon because I can draw a line on this log graph of "effective computing power" (whatever that is). AI labs need to think more about how the national security state is going to get involved. Thanks for all the help Holden"

The summary is basic received

anon_zutu said in #2005 14mo ago: received

I felt the same after reading the text, which is why I wanted to open a discussion. The tech scene has been taken over by grifters, and I the hype with AI is taking a more annoying turn than usual because it is actually more potentially useful for some things.

I felt the same afte received

anon_nafo said in #2039 14mo ago: received

>>1931
> It's true that silicon valley isn't taking info security seriously.

If there is physical defencetech, there must also be intellectual defencetech, and we are free to get our technology from external sources as needed. There are some points of particular interest to Greater San Francisco.

There are two famous universities in the chinese northern capital, Peking University (PKU) and Tsinghua (THU). The one relevant to Silicon Valley is Tsinghua University, which has economic and technological dominance. Established on the site of a Ts'ing (Qing) royal garden near the Old Summer Palace, it has become a preferred destination for western tech envoys to the Celestial Empire, including Bill Gates, Tim Cook, and Mark Zuckerberg. Mark Zuckerberg, in particular, learned chinese (pekingnese) and emphasized that his wife was chinese when hoping to obtain access to the chinese market. I note that Priscilla Chan has ancestry from the Pearl River Delta, as does much of Little Toishan in San Francisco.

Google recently added Cantonese to Google Translate, and it outperforms Baidu's and Bing's. I'm pleased that they understand that Shumchun, as an industrial centre in the Pearl River Delta, is an inalienable part of its inviolable territory.

Now, why is China's THU so important to industry? China's Silicon Valley (Zhongguancun), is located near THU. Of particular importance is Tsinghua Unigroup, a Chinese state-owned semiconductor manufacturing firm. In Taiwan, the analogous site of Hsinchu Science Park is located near the ROC versions of Tsinghua and Chiao Tung University. Our Silicon Valley has Berkeley and Stanford.

You might be wondering what happened to PKU down the street nearby the Old Summer Palace. Some of the most recent leaders of China, prominently including Xi Jinping, Hu Jintao, Zhu Rongji, Wu Bangguo, Huang Ju and Ma Kai are all THU alumni, and one would expect them to favor THU over PKU, which they do. THU alumni are found all over the world, including Hsinchu Science Park and our own Silicon Valley, giving back generously to their alma mater. At present, Hsinchu's Science Park and Silicon Valley are no longer as interesting as Cerebral Valley in San Francisco itself. We should learn from how China's elites treat the strategic importance of their educational institutions far more seriously than we do.

If there is physical received

anon_xogu said in #2041 14mo ago: received

>>2039
I don't know what you're saying but it sounds interesting.

I don't know what yo received

anon_ruta said in #2044 14mo ago: received

If drones are to be imported into the US, they had better be obtained from Shumchun, Cantonia, and not Shenzhen, Guangdong Province, People's Republic of China.

If drones are to be received

anon_gimu said in #2052 14mo ago: received

If any of you anons are in San Francisco, feel free to come to this Jungle Asian party happening on Friday:

July 12 marks the 8th anniversary of the 2016 South China Sea Arbitration case. We will join our Southeast Asian allies in San Francisco to protest at the Chinese consulate against the CCP's disregard for the ruling of the Tribunal and international order, and its continued use of the occupied Cantonia as a base to invade the territorial waters of the Philippines and Vietnam!

2024-07-12, Friday
Chinese Consulate
1450 Laguna St, San Francisco
1130 - meeting
1200 - rally starts

Anti-China Expansion Movement, Norcal Vietnamese Community, US Pinoys for Good Governance, Norcal HK Club, US Hongkongers Club, Filipino-American Human Rights Alliance, National Youth Movement for the West Phillipine Sea, Cantonia Independence Party, Yunnan Independence Party, Goetland Independence Party, Hokkienam Independence Party.

If any of you anons received

anon_gimu said in #2053 14mo ago: received

Rally name: End The Occupation, China Out Now!

Rally name: End The received

You must login to post.