The "water sellers" behind the giant AI melee have made numbness

Following the release of Wen Xin's words and Microsoft's full access to GPT-4, the big guys have new actions.

At 11 p.m. on March 21, GTC 2023 opened, NVIDIA CEO Jensen Huang delivered a speech, mentioning "AI ushered in the iPhone moment" four times in 78 minutes, causing an uproar. In addition, Huang Jenxun also announced new products and a series of subsequent actions in cooperation with Amazon, Microsoft, and Google. Then on the morning of the 22nd, Adobe and NVIDIA released new tools for using artificial intelligence to generate images, and solved the two most controversial legal issues - copyright and payment.

After listening, I can only call out, NVIDIA, as the supplier of the giants' AI arms competition, really wins.

Last month, there was a factual error in the debut demo, resulting in Alphabet's market value loss of $100 billion Bard also "made a comeback", officially released on the morning of the 22nd, Google bluntly said that this move is to widely obtain feedback from users to support its competition with Microsoft on the conversational generative model track.

According to reports, Bard is an experimental conversational AI service powered by Google's LaMDA model that can access and process information from the real world in real time through Google Search to provide up-to-date, high-quality responses. Google CEO Sundar Pichai tweeted that the open use will start with users in the United States and the United Kingdom.

After Tencent announced its latest financial report on the 22nd, Tencent Chairman and CEO Ma Huateng, President Martin Lau and other management held a conference call, at which Tencent management confirmed for the first time in public that it would launch Chat GPT-like chatbot products.

The giants are moving frequently, and the smell of gunsmoke around generative AI is getting stronger and stronger.

Giant dogfight

On the morning of the 22nd, Google announced the official public release of its chatbot product Bard. This open use will start with users in the United States and the United Kingdom, and currently only supports English and does not have coding capabilities, so responses to the code are not supported.

Google, which previously planted a head on Bard, is obviously more cautious in this release. Not only has the quota changed from an all-open to a gradual opening of the application system, but there is also a line of disclaimer below the text box and under the answer generation box: "Bard may generate inaccurate or slightly offensive information, but it does not represent the official views of Google." At the same time, there is also a large experiment logo next to the Bard logo on the page, indicating that it is still in the experimental stage.

Krawczyk said: "We know the limitations of this technology, so we are very cautious about the speed at which we can roll it out. ”

Notably, Google positions Bard as "a complement to the search engine, not a search." "Maybe it's to differentiate itself from Microsoft's search engine, New Bing."

Compared with the cautious Google, Microsoft is much more aggressive.

Last year, Microsoft took the lead in the ChatGPT wave by joining forces with OpenAI, and its stock price soared. Since the beginning of this year, the company's stock price has risen by more than 15%. As of March 23, the stock price reached $264.680 per share, and the market capitalization soared to $2.03 trillion. By comparison, while Google also gained more than 13%. But after the online demo Bard rollover in February, Google lost $100 billion in market value.

Microsoft also officially announced last week that its commercial software Microsoft 365 will be fully connected to GPT-4 and launched a new Copilot, which has gained a wave of attention. In fact, just a day ago, Google also announced that it would integrate its AI tools into the office system Workspace, but it was far less concerned than Microsoft. Since Microsoft launched its version of AI on Feb. 7, New Bing's page visits have increased by 15.8%, while visits to Google's search engine have dropped by nearly 1 percentage point.

Looking back at this time, the two companies are basically back and forth, the release of similar new products is overwhelming, and now the war between the two has burned from search and browser all the way to the field of office software.

The giants are hot, and NVIDIA is full of ambition and wants to save the bow.

At the beginning of this year, the global explosion of ChatGPT caused NVIDIA's stock price to soar, and the market value directly increased by more than $80 billion. Currently, NVIDIA has a market capitalization of $653.8 billion.

At this conference, Huang Jenxun expressed his confidence in generative AI, and has made new moves in various aspects such as quantum computers, lithography machines, and cloud computing, hoping to "empower AI companies to accelerate production."

Among them, DGX Cloud was highlighted. In terms of cloud services, NVIDIA has already cooperated with Oracle, and then Microsoft Azure and Google Cloud will also begin to host DGX cloud. Hosting DGX cloud services to businesses willing to build new products and develop AI strategies means customers can connect NVIDIA DGX AI supercomputers to the company in real time through a single browser.

Huang said the partnership brings NVIDIA's ecosystem into the hands of cloud service providers while expanding NVIDIA's market size and reach. Enterprises will be able to rent DGX cloud clusters on a monthly basis, ensuring they can quickly and easily scale large multi-node AI training.

Then on the morning of the 22nd, Adobe and NVIDIA released a new tool "Firefly" that uses artificial intelligence to generate images, allowing users to describe the images, illustrations or videos that their software will create through words. It also resolved copyright and payment, the two most controversial legal issues before.

In contrast, when Wen Xin's words were released, Baidu's Hong Kong stock fell nearly 10% and then narrowed to 6.36%.

In this regard, Li Hongyi said in an interview with the media on the 22nd: Now (the stock price falls) is that everyone has no opportunity to test Baidu's products, so this reaction may be caused by a panic about the unknown. When they get the product and have the opportunity to try it themselves, I believe that the stock price will come back.

Since then, his claims have also been confirmed, and as of the close of trading on March 23, Baidu's Hong Kong stock has rebounded, rising to HK$150.900 per share.

AI ushered in the iPhone moment

"Now one person plus AI can start a startup, and AI supercomputers can augment all the algorithms and data. That's quite a shift, and a lot of what we're doing right now is built on super-AI computers and large language models, which is a watershed moment in AI. Jensen Huang said at the GTC conference. And he also described the 2023 time node as "iPhone moment." ”

Over the past time, there has been a lot of discussion about ChatGPT and AIGC. Technology, demand, and the preconditions of the industrial chain have gradually grown to a critical point, so this long-existing thing has been pushed into a broader vision.

Whether it is Bill Gates' "the significance of the emergence of ChatGPT is no less than the birth of the Internet and personal computers", or Musk's argument that "we are not far from artificial intelligence that is so powerful that it is dangerous", it is obvious that it is placed in an extremely high position.

On the 21st, Bill Gates spoke out about ChatGPT and its related technologies, saying in his blog post "The Age of Artificial Intelligence Has Begin" that OpenAI's GPT AI model is the most revolutionary technological advancement he has seen since he first saw a graphical user interface in 1980.

"I know I've just seen the most important technological advancement since graphical user interfaces," Gates said, "and the development of artificial intelligence is as important as the birth of microprocessors, personal computers, the Internet, and mobile phones." It will change the way people work, study, travel, access healthcare, and communicate. The entire industry will be redefined around it, and businesses will be divided by how they use GPT. ”

And the technology is iterating faster and faster, which also accelerates the sense of urgency of all parties, take GPT-4, OpenAI has said that they completed model training in August 2022, and have been iterative optimization since then. Just six months later, GPT-4 was born, and all capabilities were greatly improved. In terms of security, compared with GPT-3.5, GPT-4 non-compliant content response requests are reduced by 82%, and the response accuracy is improved by 40%. In fact, GPT-4 scored 40% higher.

However, as for whether China will produce another OpenAI, an organization that builds ChatGPT, Li Yanhong said in his response to 36Kr that it will basically not. He said that OpenAI was born because the big American manufacturers are not optimistic about this direction, but now the big Chinese manufacturers are optimistic about the AI big model and are doing this direction. It doesn't really make much sense for startups to redo a ChatGPT, there are great opportunities to develop applications based on large language models, and there is no need to reinvent the wheel.

Li Yanhong believes that in the era of big models, the biggest entrepreneurial opportunity is in application. In the era of mobile Internet, there are actually few operating systems, and the most successful are WeChat, Douyin, Taobao and other applications. In the next decade, opportunities may be born that are ten times worth as many as these applications.

At present, ChatGPT has been applied in various scenarios such as content creation, office assistance, and game products, and will also empower intelligent driving, humanoid robots and other fields in the future, with broad prospects in both To B and To C. Gartner predicts that AIGC will account for 10% of all generated data by 2025. According to Frost & Sullivan and Toubao Research Institute, China's conversational AI market was 8.27 billion yuan in 2021 and is expected to grow to 26.58 billion yuan by 2026, with a CAGR of 26.3%.

Ren Zhengfei has also publicly stated that he should pay more attention to applications, saying that "the application of models sometimes has more prospects than the model itself." ”