Source: Titanium Media
Author: Lin Zhijia

ChatGPT, a big model application for artificial intelligence (AI) conversations, was released more than 4 months ago, but the AI wave it brought about is still spreading frantically around the world.

March of this year can be called “Generative AI (AIGC) Collective Outbreak Month.” Within two weeks, GPT-4, Microsoft Microsoft 365 Copilot, Midjourney V5, Google Palm API, Baidu's “Wenxin One Word”, and new Nvidia graphics cards appeared one after another. Business celebrities such as Innovation Factory CEO Li Kaifu, Meituan co-founder Wang Huiwen, Ali Vice President Jia Yangqing, and Sogou founder Wang Xiaochuan plan to end up incubating big AI models.

On March 24, the US company OpenAI also launched the ChatGPT plugins plug-in function. The first batch of 15 plug-ins was tested to achieve real-time information feedback on the Internet, which is thought to illuminate “AI's App Store App Store Moment.” According to reports, based on the estimate that the Apple App Store has paid 320 billion US dollars in revenue to developers, ChatGPT is expected to drive the technology developer market on a scale of 100 billion dollars.

Moreover, under the ChatGPT craze, just half a year after the new model was released, an American AIGC company Character.AI, which was founded 16 months ago, received round A financing led by A16z, with a valuation of over 6.8 billion yuan to become an AI unicorn. At the same time, the OpenAI Startup Fund recently led the financing of Norwegian 1X humanoid robot company; after “Wen Xin's words” was released, due to insufficient computing power, Baidu purchased Nvidia chips in large quantities at a premium of more than 50%...

“I have a deep feeling that ChatGPT is already making all entrepreneurs anxious.” An AI industry entrepreneur told Titanium Media App that investors are persuading him to switch careers and invest in big AI models, but lack of talent, lack of large chip computing power, and tight liquidity are still important bottlenecks in the company's transformation. He believes that the continuous hype of investment institutions and markets has forced some entrepreneurs to join the “inner volume” of the big model field.

The 100 billion “banknote” capability that has emerged in ChatGPT has not only sparked big discussions from investors to entrepreneurs, from unicorns to big manufacturers, from industry to academia, from economists to ministers of science and technology, but is gradually evolving into a crazy, anxious, and infatuated competition.

Zhang Shu (pseudonym), CEO of a leading AI company, said bluntly to the Titanium Media App:“This wave of big models will definitely 'drop in one place', because the technology thing really isn't a 'great miracle. ' I think there will be a lot of 'bubbles' in the AI industry in five or six months.”

A crazy, anxious, and insipid money ability competition

Some time ago, the Titanium Media App asked Dai Yusen, one of the investors in Wang Huiwen's newly founded company outside of Lightyear and managing partner of Zhenge Fund:Is Zhenge really sure that the next “Perfect Diary” (Yixian E-Commerce, YSG) will be cast in this AIGC wave?

And Dai Yusen's answer was: How do you know if you don't try it?

This has become a microcosm of the development of the industry under the current ChatGPT explosion. Everyone is “betting” on the future in uncertainty:A group of investors who gamble like “Sun Zhengyi”, many business celebrities full of confidence, and some anxious existing AI entrepreneurs and technology “migrant workers” collectively joined this “inner volume” vortex of ChatGPT, AIGC, and Big Model.Furthermore, the iteration speed of AI big model messages measured in “days” made everyone more “anxious.”

Let's review the timeline of this round of ChatGPT and the big model craze:

• 2012 was the first year of deep learning. With the introduction of deep learning algorithms into ImageNet as an important node, AlexNet successfully reduced the image recognition error rate by 10.8 pcts through the construction of the deep neural network AlexNet, which proved the development potential of deep learning; in 2017, Google released the Big Model (Transformer) technology route, which would reduce marginal costs, which began to attract the attention of the industry and domestic and foreign giants, and gradually became a consensus for development;

• In 2022, OpenAI launched Dall·E2 to achieve the creation of literary images and increase the resolution;

• After ChatGPT was released on November 30 last year, the number of registered users exceeded 1 million in just 5 days, and after 60 days, there were more than 100 million monthly active users;

• In February of this year, Microsoft announced a multi-billion dollar investment in OpenAI. The latter's valuation reached 29 billion US dollars, setting a new high in single financing for the AIGC industry;

• In the past two months, Chinese and American tech giants such as Google, Ali, Baidu, and Adobe have successively announced big models or ChatGPT-like technologies, which are applied to fields such as text and image generation.

As a result, many companies, investors, and academic experts intensively discussed topics such as the impact and investment opportunities brought about by ChatGPT.

Minister of Science and Technology Wang Zhigang said on March 5 that ChatGPT proved that AI is the general direction, and that OpenAI has an advantage in the real-time effects of AI dialogue. “For example, engines, everyone can make engines, but the quality is different. Soccer is all about putting up and shooting, but it's not easy to be as good as Messi.”

So why is society experiencing insanity, anxiety, and internal warfare over ChatGPT? In response to this, the Titanium Media App entered many prompts to ask GPT-4.

GPT-4's answer is that in the past four months, large-scale model “banknote” capabilities have been concentrated, and the reason for a crazy, anxious, and internal war has arisen, mainly stemming from technical, commercial, and talent competition.

On the technical side,ChatGPT is an advanced technology in the field of natural language processing, and many companies and organizations want to lead the way in this field. This has led to fierce technology competition, and enterprises are continuously improving their technical level and R&D investment to gain more market share.

On the commercial side,ChatGPT has broad application prospects in many industries, such as intelligent customer service, intelligent translation, automated writing, etc. This has led to fierce commercial competition. Companies and organizations compete for ChatGPT's commercial application market, hoping to obtain a larger share of the market and higher profits.

In terms of talent,The development and application of ChatGPT requires extensive talent support, including natural language processing experts, algorithm engineers, product managers, etc. As a result, companies and organizations are also fiercely competing to recruit, develop, and retain these talents. At the same time, career uncertainty brought about by AI has caused anxiety. OpenAI's official research estimates that for 19% of jobs in the US, at least 50% of the job content will be affected, and for 80% of jobs, at least 10% of the job content will be affected to a greater or lesser extent. Mathematicians, accountants and auditors, news analysts, legal secretaries and administrative assistants, tax reporters, etc. are the most susceptible to GPT.

Furthermore,Capital involvement may cause hype in the natural language processing and AIGC technology fields where ChatGPT is located. GPT-4 believes that some companies, investment institutions, and individuals may use media publicity, marketing, etc. to hype up ChatGPT and other technologies, thereby obtaining more investment and profit.

The Titanium Media App discovered through the Baidu Index that February 8 was the Internet information explosion point in this round of the ChatGPT boom. Baidu's ChatGPT keywords increased 1755% month-on-month. Judging from the incident, in the early morning of February 8, Microsoft announced the launch of the ChatGPT version of the new Bing (Bing) search and Edge browser. Google and Baidu later announced the launch of a ChatGPT model application; the Alibaba version of the ChatGPT message was reported on February 8, and related functions were directly integrated into DingTalk customization.

Why is the big model exploding at this moment? “Simply put, I don't know, and I don't know which month is the most critical; suddenly I have the ability. But I'm sure future humans will figure out the underlying theory.” Li Yanhong, founder and CEO of Baidu, said that the current theory cannot be explained. In fact, none of the big companies are working on generative AI, and they haven't invested too many resources in it, including academics. No one thinks this is worth studying, but once it comes out, it's really amazing and gets a lot of attention.

Today, the crazy, anxious, and internal competition sparked by ChatGPT continues.

An industry source told the Titanium Media App that since big models require huge computing power support, Baidu currently lacks chips. It is purchasing large quantities of Nvidia A100/A800 chips and server cards for ChatGPT products such as ChatGPT at a premium of 50% or more. “Baidu now says how many will be needed. The Nvidia server chips purchased by Baidu last year are only being shipped now. The spot premium is very high because Nvidia is also rising in price.” The person mentioned above said.

Ma Yun talks about ChatGPT. This type of technology is only the beginning of the AI era

Today, ChatGPT is more than just a conversational model app.

On March 24, Beijing time, OpenAI announced that ChatGPT will launch plug-in tests. The first batch of plug-ins included the three apps officially released by OpenAI and 12 apps collaborated with third parties. OpenAI said that these plug-ins will first be open to paid users, while developers can also apply for a waiting list to develop plug-ins.

“Since we released ChatGPT, users have hoped to be able to provide plug-in functions to unlock a large number of application possibilities. We will start with a small number of requirements and gradually expand more applications. We can't wait to build a community that provides models for future human-computer interaction,” OpenAI said in the developer log.

In fact, OpenAI introduced a plug-in function that can solve the problem of poor data feeding in ChatGPT, directly connect to global Internet searches, such as the Chinese Wikipedia database, and embed apps and applets within ChatGPT to form an application store or operating system. It also represents that ChatGPT has officially entered the AI world and can have the most direct connection with people's daily work.

Nvidia AI scientist Jim Fan said in this regard that if the launch of ChatGPT was the “advent of the iPhone,” then today (the launch of the plug-in function) is the “iOS App Store” event. He also said that the “App Store” is just the beginning, that OpenAI's master plan is to make “an app that has everything,” and that day will come soon. He believes that the addition of plug-in functions to the ChatGPT ecosystem has become a metaverse-level operating system.

Li Changliang, former vice president and head of the artificial intelligence division of Jinshan Software and a current entrepreneur in the AI field, said that the advent of the OpenAI plug-in will definitely bring developers the same opportunities to make money in the early days of Apple, and even “greater opportunities” than at that time.

A new era of AI productivity has quietly arrived.

So, under the ChatGPT and AI 2.0 frenzy, where are the entrepreneurial and investment opportunities for future technological change?

The Titanium Media App discovered that OpenAI established a 100 million US dollar startup fund, the OpenAI Startup Fund last year, and the accelerator program Converge, which included Microsoft and other organizations acting as LPs to lay out to early-stage AI startups. Although the fund is small, OpenAI has made more than a dozen investments so far. The latest project is the Norwegian 1X humanoid robot company mentioned at the beginning.

The Titanium Media App is concerned that most of the projects invested in the OpenAI Venture Capital Fund are AIGC creative applications, chip technology, integrated development environments (IDE), databases, and AI tool applications, and most of them are large-scale model applications.

Therefore, scenario application of big models may be a top priority in future technological development.

Li Yanhong said that there will be brand new entrepreneurial opportunities on the big model application layer, ten times that of WeChat and Douyin now. There are many opportunities at the application level, and the value that can be created is also huge. Moreover, he believes AI will disrupt the cloud computing market.

Ma Yun also talked about ChatGPT. According to Yungu Education Officer WeChat, Alibaba founder Ma Yun said on March 27 that technologies such as ChatGPT have already posed challenges to education, but ChatGPT technology is only the beginning of the AI era. We need to use AI to solve (apply) problems rather than be controlled by AI. Although human physical and mental strength is no match for machines, machines only have a “core,” and people have a “heart.” According to reports, Yungu School is a private school invested and founded by Ali's partner.

Editor/jayden