News on China's scientific and technological development.
- Thread starter Quickie
- Start date
Didn't some university in china created a model called WuDao 2.0 but they didn't commercialised it?
This is interesting. Unisoc along with China telecom satellite corporation, ZTE, Baicabang, China Academy of information and Communications Technology, China Telecom Guangdong Company and other industrial partners, successfully completed the world's first S-band¹ 5G NTN (non-terrestrial network, non-terrestrial network) network) technical star verification.
satellite 5g?
It's difficult for two reasons:
- Unfortunately, English is the global language, and as such there is much more media, text, etc. in English than there are in Chinese. The GPT-3.5 model, which scrapes the internet for this material, therefore has much access to a larger and more diverse set of training data. This is the cultural advantage the West has built up by the last 400 years of colonialism, which is difficult to overcome.
- While academically, China is just as productive as the West in terms of AI research output, Western AI companies are still generally ahead of China's. This is because large corporations - like Baidu, Google, Tencent, Microsoft, etc. - are not well suited to the task of making breakthrough bets, being tied to their investors' short term interests. This is also why Google sat on similar technology to Open AI for years, but did nothing with it. It took a start up like Open AI to revolutionize the NLP industry; just like it took Stability AI and Open AI to revolutionize the AI art industry (even though, again, Google already had technology like this with Imagen).
China needs to put together similar companies if it wants to compete in this space, because even though Chinese AI researchers are doing very well vs. the West in the academic world, in the commercial industry, Chinese companies have yet to realize its potential. To be fair, neither have most large Western corporations and there's a reason for that.
Nonsense. Most of the underlying tech that OpenAI relies on was in fact pioneered by Big Tech companies such as Google, Meta etc. So if Big Tech companies didn't want to make breakthrough bets, how come they invented the stuff? It's a self-refuting argument.
They sat on the tech for other reasons (fearing bad PR, excessive regulatory oversight) which OpenAI didn't have to worry about since they are small. That's why they released ChatGPT. Being small has its upsides. Well, I suspect that may now also slowly be changing as they are getting more and more integrated into MS.
P.S. Tsinghua has an LLM model very similar to ChatGPT's.
That's the correct assumption, word on the Silicon Valley grapevine is that Google and other big tech have several times refused to buy OpenAI, they're all letting a small company test the waters and explore controversial use cases first before jumping in. A lot of the lucrative use cases can disrupt entire industries and incur huge public/legal backlash.
supercat
Major
Actually, there are two similar AI bots under development in China. 360 Security Technology's version will be released ASAP, while Baidu's version will finish internal testing in March. I doubt the Chinese language-computer interference is a problem at all with today's technologies.
They published papers and built prototypes, while Open AI built an actual product that's revolutionizing the industry, which they are now playing "fast follower" on. That's the difference.
I recognized in that very post that Google, Microsoft, etc. have the *capability* to develop the technology but don't, for various reasons. The biggest of which is that, as large, public corporations, they aren't willing to invest in high risk products, choosing instead to focus on sure bets.
But with that sort of mentality, you will never be first, and being first is important due to first movers' advantage. Google may still be able to over take Open AI because of its natural platform advantages (Android, Google Search), but if it was any other company, it'd have been game over as ChatGPT is already in production and hundreds of millions are already using it. This is why Google declared "code red" the moment ChatGPT released. They knew how disruptive a technology it will prove to be and that if they don't match within a short amount of time, they'll lose the market.
As China, you want to be on top of these AI technologies. Not just a "fast follower," but a leader, like Byte Dance was in the social media space.
For that, you'll need start ups with a lot of energy and ambition. Fortunately, China has seen the rise of such start ups in other fields (electrical vehicles, social media, mobile games). So there's hope it'll manage in the AI space, too, though the US is doing its best to stop it (by trying to ban Western investors from investing in Chinese technology start ups).
China needs to grow and develop it's own GMO crops and livestock. This is an extremely disruptive tech that has tons of potential but China doesn't allow for the growing of GMO crops on it's own soil and thus most chinese companies aren't investing heavily in this area. This is despite the massive threat of having it's food supply disrupted or falling behind on core GMO technology. Or the fact that they're already buying tons of GMO crops from overseas anyway.
I know that China has had great success with cross breeding plant species, like with their salt tolerant rice, high yield hybrid rice and perennial rice, but those methods were outdated 20 years ago, it takes decades, and they're still using other outdated methods like growing crops from seeds that were in outer space, in hopes that some random mutation will give them a good strain. CRISPR allows you to actually target whatever gene you want and it takes a fraction of the time and resources.
Existing GMO crops already give a 5-10% increase in yield and the world is barely scratching the surface.
There's the current GMO crops that's changing just a handful of genes. But the real breakthrough is when you're making radically different changes to hundreds of genes. Just look at C4 rice project, it could give the world a rice strain that has 50% more yield, needs 50% less water and much less fertilizer. And you can use that knowledge to help convert even more C3 plants to C4. Or even other more lucidious ideas. Like a rice strain that has just as much protein as meat. Converting staple crops to high yield perennial strains. Livestock that's immune to most of the diseases currently in circulation and grows fast and fat without the need for anti-biotics.
This is something that China should have done 20 years ago, let alone in 2024-2025.
I know that China has had great success with cross breeding plant species, like with their salt tolerant rice, high yield hybrid rice and perennial rice, but those methods were outdated 20 years ago, it takes decades, and they're still using other outdated methods like growing crops from seeds that were in outer space, in hopes that some random mutation will give them a good strain. CRISPR allows you to actually target whatever gene you want and it takes a fraction of the time and resources.
Existing GMO crops already give a 5-10% increase in yield and the world is barely scratching the surface.
There's the current GMO crops that's changing just a handful of genes. But the real breakthrough is when you're making radically different changes to hundreds of genes. Just look at C4 rice project, it could give the world a rice strain that has 50% more yield, needs 50% less water and much less fertilizer. And you can use that knowledge to help convert even more C3 plants to C4. Or even other more lucidious ideas. Like a rice strain that has just as much protein as meat. Converting staple crops to high yield perennial strains. Livestock that's immune to most of the diseases currently in circulation and grows fast and fat without the need for anti-biotics.
This is something that China should have done 20 years ago, let alone in 2024-2025.