The United States has quietly achieved a breakthrough
AI entrepreneurs sigh: The United States has quietly achieved a breakthrough
The impact of chatGPT will be far-reaching. There is a gap between China and the United States in terms of algorithms and hardware. However, this gap cannot be bridged in the short term.
Check out some amazing recent uses of chatGPT.
For example, chatGPT broke through the programmer test of major manufacturers. chatGPT obtains a doctor's qualification certificate. chatGPT wrote the best philosophy papers in the class, so that the school banned the use of chatGPT to write papers, etc., and chatGPT has just been public beta.
Imagine if there is a person who is a professional lawyer, a professional doctor, an investment consultant, a general doctoral supervisor, a general teacher, a psychological counselor, and even a poet, web writer, etc. Then you can find him to solve all kinds of problems in your life. Do you still need to seek advice from doctors, lawyers and other professionals? Your life is more convenient, but will the demand for these jobs become less and less?
People used to search on search engines before seeking out professionals. Using chatGPT is equivalent to accessing professionals as conveniently as using a search engine.
In addition to search engines, the biggest impact may be the service industry. This is not the same as we expected before. It was previously expected that manual workers would be the first to be replaced, but it turned out that white-collar workers were the first to be replaced.
Musk said that AI will be more terrifying than nuclear bombs. Now, he really came.
Huang Renxun commented on chatGPT: It is equivalent to the advent of the iPhone.
What you said is true. chatGPT uses massive data on the Internet for training. Coupled with a lot of interactive communication with real people, it can actually make chatGPT continue to optimize.
With the fire of chatGPT. Domestic companies have announced that they have launched or will start research immediately. However, according to the entrepreneur, most domestic companies use open-source algorithms (it will no longer be open-source after GPT3), and because of insufficient training data, insufficient computing power, etc., the effect may be greatly reduced, or even troublesome. Make a lot of jokes.
A recent piece of news is that the Google version of ChatGPT is not satisfactory, and its market value evaporated by more than 700 billion overnight. Compared with Google's computing power, there is no company in China that can compare with it.
Now looking at the United States getting stuck on the chip, it is obvious that the ulterior motives are not drinking, and its real purpose may be AI.
The most critical aspects of AI are three aspects, data, computing power, and algorithms.
If the chip is stuck, the computing power is stuck. The purpose of the United States in terms of chips is very clear. If the computing power of American companies' chips exceeds a certain value, they will not be allowed to sell to China. If the chip design of a Chinese company exceeds this value, then TSMC, Samsung and other companies are not allowed to manufacture it. It even directly blacklisted outstanding domestic AI companies. Therefore, domestic companies have basically not mentioned computing power recently, mainly because they do not want to cause trouble.
In terms of algorithm, chatGPT was originally open source, but it is no longer open source after GPT3 version. Even the registration of the openAI website excludes China as well as Russia and Iran.
Data should be where we have an advantage. However, various domestic websites have created small circles and banned search engine access, resulting in many isolated islands of information. Each company operates in its own way, unable to integrate data, and it is difficult to take advantage of the large amount of data.
In terms of AI, especially big data models, we have fallen behind. The United States made a major breakthrough secretly.
I personally think that chatGPT can appear in the United States, which shows that the innovation ability of the United States is still the same as before. In the competition between China and the United States, in addition to the competition of scale, what is more important is the ability to innovate.
In order to innovate, the first thing is talents. Domestic training is one aspect, and the other is to attract talents from abroad. Google founder Sergey Brin was born in Russia. Musk, the boss of openAI and the boss of many famous companies, was born in South Africa. The inventor of chatGPT is also a Russian. Nvidia's boss is from Taiwan, and so on.
Looking forward, the rise of the United States in World War I and World War II was not also achieved by a large number of European scientists going to the United States for scientific research and exploration?
The United States gathers talents from all over the world to work for it, so we should not limit the recruitment of talents to Chinese and local experts. We should continue to improve the research and development environment to attract and encourage those dreamers who want to change the world to come to China to realize their dreams. . We are a multi-ethnic country, and we should be more inclusive.
Domestic entrepreneurs should also take responsibility. There are many outstanding entrepreneurs in China, such as Ren Zhengfei, Cao Dewang, Wang Chuanfu, etc., who have also made world-renowned contributions in their respective fields, but generally speaking, there is still a lack of entrepreneurs with more lofty thinking like Jobs and Musk. In particular, a few well-known entrepreneurs still only think about making money after making a fortune, and then want to change their nationality after getting money. Their ideological realm is really not flattering.
In the current direction of AI, although the United States has gained the first opportunity, we are not without hope.
The level of domestic AI, it should be said, is not far behind. Many major manufacturers, including Huawei, have already researched big data models simultaneously. At the same time, we also have our own advantages, such as electricity.
One of the main reasons why we can come from behind in the electric vehicle industry is that our electricity is cheap. The high-voltage power grid is world-leading.
And chatGPT itself is also a big energy consumer. UBS estimates that chatPGT uses 10,000 Nvidia GPUs, and an Nvidia A100 GPU has a minimum power consumption of 46W and a maximum power consumption of 407w. Thinking about more than one data center based on chatGPT, this requires a very strong power grid and sufficient and stable energy supply.
We are behind in AI, but don't forget that the iPhone was born in the United States, but we also have Huawei, Xiaomi, oppo, vivo, etc. Tesla was born in the United States, and we also have BYD, Weilai, Xiaopeng, etc. Now that chatGPT is born in the United States, who will China's chatGPT get?