Artificial Intelligence thread

[Overbom]

I hope the Chinese government and corporations doesn't wait till confirmation before they start pushing harder toward general AI. This will have as much, if not greater, impact then the first industrial revolution. The comparative differences between countries that have it and don't have it is like one using modern day calculator to the one using abacus. Yes, there are many groups in China are making new models and model with higher parameters and stuff but it seems like there is a lack of leaders/visionaries which can harness and use all these tools to produce globally impactful/leading products.

Rumours and expectations are that GPT-5 (released end of next year?) is going be a very big improvement over GPT-4.

GPT-4 is already impressive enough to have commercial applications. If the GPT-5 proves to be as good as people expect it to be, it is going to be a huge advancement with many implications for the world.

 

[mossen]

Baidu has already admitted that the sanctions will put a dampener on their AI development. Most AI progress will remain concentrated in the US for the time being. China a distant second and Rest of World basically out of the race completely.

 

[Overbom]

Baidu has already admitted that the sanctions will put a dampener on their AI development. Most AI progress will remain concentrated in the US for the time being. China a distant second and Rest of World basically out of the race completely.

"Distant"..

Just the Yi 30b model from 01.Ai (Chinese start up) obliterates ChatGPT.

Open Source models in general are approaching GPT-4 level.
Although OpenAi is undoubtedly on the forefront, I would say that the gap is at least 6 months and at most 1 year.

The big issue is of course future development as sanctions on Nvidia AI cards are going to slow down training of Chinese LLMs. However I don't think the slow down will be severe, at most a few months extra. Hopefully Huawei can provide competitive alternatives with its Ascend product lineup.

Don't forget that although Nvidia is more advanced on chip node, Huawei has huge advantages on networking. And on massive AI usage, communication between each card is the biggest bottleneck.

 

[tphuang]

Baidu has already admitted that the sanctions will put a dampener on their AI development. Most AI progress will remain concentrated in the US for the time being. China a distant second and Rest of World basically out of the race completely.

where did you get this from? It says that it has enough Nvidia chips for next 2 yrs and we all know it is already sourcing Ascend chips.

The problem in China is there are too many companies rushing after the LLM gold mine. too many players are buying up chips to be the next OpenAI. Now, they may not have as much computational resource or AI talent as openAI, but they will inevitably be adapted in more areas.

So how exactly did you reach the conclusion that China is a distant second? have you thought about how many AI areas are there and who are the leaders in each areas?

 

[ougoah]

where did you get this from? It says that it has enough Nvidia chips for next 2 yrs and we all know it is already sourcing Ascend chips.

The problem in China is there are too many companies rushing after the LLM gold mine. too many players are buying up chips to be the next OpenAI. Now, they may not have as much computational resource or AI talent as openAI, but they will inevitably be adapted in more areas.

So how exactly did you reach the conclusion that China is a distant second? have you thought about how many AI areas are there and who are the leaders in each areas?

Would be nice for this thread and those in the field or parallel fields with deeper knowledge on the topic to provide the thread or forum with a good summary of the major industry players in China, their specialties, what they've achieved, the general performance specs and parameters of their offerings/products and what they're working on. Obviously, publicly accessible knowledge and as above board as possible.

 

[tacoburger]

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

Self-driving company Pony.ai has made another breakthrough in China with the granting of a license to conduct “platooning” tests of its autonomous trucks on public roads.
The approval has come from the municipal government of the mega city of Guangzhou – where Pony.ai has a base – in the southern province of Guangdong. Platooning is when a number of vehicles travel in convoy in a designated formation, which brings several benefits.
Essentially what this means is that one autonomous truck will lead multiple others in a line as they drive along highways. As five vehicles have initially been awarded permits, this means that four, in theory, could follow the leader, although images released by Pony.ai so far show just three trucks – in a “1 + 2” formation – out testing.
According to Pony.ai, the advantages of platooning include reduced operating costs and improved efficiency.
As a post on the company’s WeChat account explained: “By eliminating the difference in braking time between front and rear vehicles and shortening the following distance, the vehicle in front can break the wind for the vehicle behind, effectively reducing air resistance, thereby reducing carbon emissions, achieving green travel, while improving road usage efficiency and alleviating road congestion.”

Pretty big deal here. And honestly even if full level 5 self driving doesn't work out, have a fleet of trucks being led by a team of human drivers is still a pretty big deal. If a team of 1-3 people can lead a convoy of 10 trucks, that's a big efficiently boost.

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

More chinese companies that are working on self driving trucks. Self driving trucks seems like they will be easier to turn into self driving then personal use cars. They are standardized, used and maintained by a company, travel a set route and they can be used in a convoy as above. As compared to a random car model that a random person can abuse and order it to drive anywhere and everywhere and it has to account for every possible factor.

Please, Log in or Register to view URLs content!

One broad transformation is that the trucking industry in China is changing from one in which individual drivers dominated, to one with fleets holding the majority share, said Gui Lingfeng, principal at Kearney Strategy Consultants.

He pointed out that five years ago, fleet operators only had about 20% of the Chinese trucking market. Today it’s at 36%, and projected to reach 75% in 2025, he said.

The companies trying to sell trucks to fleet operators are including driver-assist tech as a way to make the vehicles more attractive, Gui said.
That early tech integration gives truck manufacturers an edge on the amount of data they can collect — for training autonomous driving algorithms, he said.
In addition, Chinese authorities require all newly manufactured trucks since 2022 to come with basic driver-assist tech for warning against forward collision and lane departure, Gui said.

Chinese driver-assist trucking startup Inceptio claims it already has more than 650 trucks operating in China — mostly for logistics customers — and covered more than 50 million kilometers (31 million miles) in commercial operations.

Other startups are testing out different forms of driver-assist trucks in China.
Kargobot, backed by ride-hailing giant Didi, operates more than 100 autonomous-driving trucks between Tianjin, near Beijing, and the northern province of Inner Mongolia.
Many of those trucks operate via what’s called platooning — having a human driver sit in the front vehicle and having two or three trucks follow behind in fully self-driving mode, with no human staffer inside.
Kargobot CEO Junqing Wei envisions that in the next decade or two, a network of hubs on the edge of cities, connected by highways on which self-driving trucks transport products. That’s according to his remarks in October at CNBC’s East Tech West conference in the Nansha district of Guangzhou, China.

Even more info on self driving trucks.

 

[Overbom]

DeepMind. Just. Incredible.
Probably even more significant than their AlphaFold research

Please, Log in or Register to view URLs content!

Millions of new materials discovered with deep learning​

AI tool GNoME finds 2.2 million new crystals, including 380,000 stable materials that could power future technologies

Today, in a

Please, Log in or Register to view URLs content!

, we share the discovery of 2.2 million new crystals – equivalent to nearly 800 years’ worth of knowledge. We introduce Graph Networks for Materials Exploration (GNoME), our new deep learning tool that dramatically increases the speed and efficiency of discovery by predicting the stability of new materials.

With GNoME, we’ve multiplied the number of technologically viable materials known to humanity. Of its 2.2 million predictions, 380,000 are the most stable, making them promising candidates for experimental synthesis. Among these candidates are materials that have the potential to develop future transformative technologies ranging from superconductors, powering supercomputers, and next-generation batteries to boost the efficiency of electric vehicles.

---

For example, 52,000 new layered compounds similar to graphene that have the potential to revolutionize electronics with the development of superconductors. Previously, about

Please, Log in or Register to view URLs content!

. We also found 528 potential lithium ion conductors, 25 times more than a

Please, Log in or Register to view URLs content!

, which could be used to improve the performance of rechargeable batteries.

 

[SDtom]

DeepMind. Just. Incredible.
Probably even more significant than their AlphaFold research

Please, Log in or Register to view URLs content!

---

This is one of the reason why I disagree with Xi Jinping's idea of focusing more on creating many Little Champions like they have in Germany. Little champions is good and all but they doesn't have the monetary and knowledge resources and visions to do what Google, a Giant Champions, does.

Like I said before, Chinese organizations, universities keep saying they are creating more powerful AI models and with billions and trillions of parameters but yet they don't have any visionaries and leaders to make use of those tools to produce leaping researches and products.

From the look of it, any large companies that hopes to be competitive globally will need to rely on AI for research and development assistance.

Little Giants get stomp on and eaten up; like how Google bought UK's Deepmine. This why China needs to create and support its own Giant Champions too.

 

[sunnymaxi]

This is one of the reason why I disagree with Xi Jinping's idea of focusing more on creating many Little Champions like they have in Germany. Little champions is good and all but they doesn't have the monetary and knowledge resources and visions to do what Google, a Giant Champions, does.

Like I said before, Chinese organizations, universities keep saying they are creating more powerful AI models and with billions and trillions of parameters but yet they don't have any visionaries and leaders to make use of those tools to produce leaping researches and products.

From the look of it, any large companies that hopes to be competitive globally will need to rely on AI for research and development assistance.

Little Giants get stomp on and eaten up; like how Google bought UK's Deepmine. This why China needs to create and support its own Giant Champions too.

i had replied..


Little giants policy has nothing to do with Giant Champions..

there are so many Ai tech companies are emerging in China. will consolidate in few years just like EVs .. many will phase out

 

[tphuang]

Please, Log in or Register to view URLs content!

Huawei's play here with this AI cloud that have all the models. everyone needs more comptuation power and AI chips are in hot demand and too expensive and short supply

no problem, use my datacenter here with pre configured models and everything, it will be more efficient.