Theuncover.ai
Region: US
Thursday 23 July 2026 15:51:39 GMT
89792
1321
108
64
Music
Download
Comments
yifan_east :
China is fighting with extra weights bounded to them 💀💀💀
2026-07-26 09:07:25
0
bubbz777777 :
As an American, I am very happy China is developing top tier AI models to compete with the U.S. It would be a dark world if only a few American models existed. They would be extremely expensive. Chinese models make sure to keep American AI companies in check and also competitive.
2026-07-23 22:39:42
138
soopersafe :
And just like eveything else, now China has the best model (kimi k3) and the gap between china and america will probably continue to grow.
2026-07-24 15:08:13
1
Milky the Cat :
am I only one that understood maybe 30% of what he said 😑
2026-07-24 02:30:31
18
War Badger :
I mean. Yes but also. There’s weird propaganda in here. Let’s take one example: Gated delta networks (linear attention) wasn’t invented by qwen it was a basic extension of taking a standard local model and assigning most of the tensors to RNN based work from guys like the RWKV team. I’m not saying it isn’t real or significant I’m just saying that it’s taking other people’s building blocks and configuring them differently. The coherent product is real, powerful, and truly innovative on edge devices for limiting kv-cache and some other behavioral optimizations over say Gemma’s attention model. But the point I’d make here is that “China” is sort of like the equivalent of IBM relative to the western AI market. They aren’t an alternative today they’re like one alternative scoped to a dozen American pillars while being given quietly national funding that say the Gemma team can’t imagine. The difference is Gemma has the institutional backing of Google with all its infrastructure, the developer pipeline around AI, and better access to hyper scaling hardware. And the point here is that this guy, who worked in both environments, is revealing that the China team has its own internal narrative as a US alternative innovator. They’re not though. They’re doing the same thing with AI that they have in other tech. They’re taking western innovation and scaling it while making the parts that they have less access to more efficient. It’s meaningful but it is truly a secondary market to the innovation in Silicon Valley. They distill off our models. They combine our architectures. They optimize and tune pre training because it’s excessively easier to hire 2k AI researchers for that task in China than 200 in America. Their achievements are real but they aren’t making iPhones. They’re copying them and making them accessible to everyone who can’t pay the Apple tax. And this is important for anyone who cares about the AI industry to understand about the institutional and geopolitical differences today.
2026-07-24 17:18:57
1
NDEREA :
What is he saying
2026-07-24 14:10:39
11
JerryluvYHVH :
If nature and biology teaches us anything. I think China will be the first to get AGI because the human brain doesn't run on unlimited computing power and memory but limited power and energy.
2026-07-24 23:32:59
6
Solomon :
Cut the crap Chinese models are distillation models
2026-07-24 07:09:35
1
Tumor :
China uses just as many if not more GPUs as America they just lie about it bc they’re not supposed to have them. Embarrassing for China to need American GPUs and America doesn’t want to give them to China
2026-07-23 18:41:56
11
jynolo_sg :
America wants to win without thinking, China thinks without the need to win
2026-07-25 02:59:21
1
XYZ :
The background music is unnecessarily dramatic
2026-07-25 04:52:41
2
听妈妈的话小瓜皮 :
美国的AI大部分也是中国AI人才帮他研发的
2026-07-25 00:32:33
2
sgthan :
I read it, but don't understand it😂
2026-07-24 17:41:47
4
dyq3ob9oobm0 :
La escasez agudiza el ingenio. Es ley de vida. Los chinos ganarán sino utilizan la misma técnica los americanos.
2026-07-25 12:38:18
2
VO :
Let’s put it this way. Either 1st or 2nd will have the greatest benefits. Because the world currently has only 2 players. Now what it’s important for both are 1. Access to stable power supply. 2. Develop an efficient chip and connectivity thru smart physics not brute force. 3. Allow as many users to access to your model. 4. Put the knowledge into use they robot, automobile, space and consumer’s application. Next, as the west famously put it; “the rest is history,…”
2026-07-25 04:26:55
0
cosmic dust :
The nutrition has to come from somewhere. If you don't have it, it does not matter if you sleep more or exercise more, that would not solve your nutrition problem. So the answer to me is obvious, no matter how you want to spin it.
2026-07-24 15:52:29
0
Dmitri :
so it’s like focusing on calories vs focusing on macros
2026-07-24 16:06:35
0
AAA :
Because none of them have a complete design of AI system. If you have a design, you know where it requires more calculating facilities, where fewer is suitable.
2026-07-24 15:01:03
0
forever Never :
Chinese models are about token efficiency? someone better tell them that...
2026-07-23 23:21:42
0
userk4m7v0nu25 🇦🇺 :
Was this funded by HR 1157 1.6B bill?
2026-07-24 09:40:53
1
Sean Gentle Poet :
Once a model becomes Open Source basically everyone can own it. TQ China. don't forget llama too. the spinoff Nemotron Ultra was really good
2026-07-24 14:28:50
2
executivecommis :
What a bunch of absolute nonsense
2026-07-24 08:05:10
0
MukbangEats :
Overglazing china’s trash tech
2026-07-24 00:29:04
0
Kenny :
Currently US media is launching systematic smear campaigns against Chinese AI models and claim distillation or intellectual property theft ideas. Many people can be so easily manipulated and brainwashed, which is in my view why the rich can be so powerful.
2026-07-24 05:02:21
8
cia nagamas :
chines as propaganda china
2026-07-25 12:57:53
0
To see more videos from user @theuncover.ai, please go to the Tikwm
homepage.
