Dream-seeking AIGC, the next breakthrough after the big model is in the application ecology

Source: Titanium Media

Author: Du Zhiqiang

It's been a hot July day, but the heat seems to have cooled down quite a bit on the big models. After the "Hundred Models War", we saw that the reaction of the investment market was not as enthusiastic as expected. In June, the decline in ChatGPT traffic also caused the AI industry to be questioned again.

There is no constant outlet, and every time a new trend comes, there is a risk of half becoming a bubble. However, this time something was different. From PGC in the Web1 era, to UGC in the Web2 era, to AIGC in the Web3 era, the big model is only a part. It is undeniable that there is a concept of speculation, but after the "big model fever" passes, it means that a new stage has begun.

For the general public, the landing application end is the place where the perception is more profound. "In the era of AIGC, every product is worth doing again." More than one industry participant expressed this view on Titanium Media App. Unlike blockchain or Metaverse, AIGC has long-term value support, and vertical landing applications have become a new round of exploration.

Shen Yang, professor and director of Metaverse Culture Lab, School of Journalism, Tsinghua University, told Titanium Media App, "The application value of the large model is still very large, it just takes a little time."

Under the new wave, the fanatical big model

The popularity of ChatGPT has sounded the horn of a new round of AI competition. Judging from the performance of the market, whether it is an entrepreneur or a giant company, the attitude given is very consistent, including Wang Huiwen, Li Kaifu and other industry veterans have also joined the battle.

Recalling the moment when the official announcement was made, Wang Huiwen couldn't hide his excitement about building China's OpenAI. He posted recruitment orders on social platforms many times, declaring that he would "accelerate AGI to benefit mankind together."

As an old partner, Wang Xing also expressed his support immediately, "The big AI model makes me excited about the huge productivity that will be created, but also worried about its impact on the whole world in the future. Lao Wang and I have walked together for nearly twenty years on the road to entrepreneurship. Since he is determined to embrace this big wave, I must support him."

After the high-profile entry of various companies, the main line of the story in the first half of the year has basically become a moment to show off their skills. While releasing the large model, they still do not forget to use their peers to improve themselves. According to incomplete statistics, as of July this year, more than 110 large-scale models have been released in China, of which more than 80 large-scale models have more than 1 billion parameters.

Among them, large factories, start-up companies, and universities have released large models. Some people in the industry complained that the large model can no longer be seen, and all kinds of names feel like relearning the vocabulary dictionary.

It is worth noting that, taking the ride of AIGC, the net worth of the CEO of the chip giant Nvidia directly doubled, and related domestic concept stocks also ushered in a double harvest in terms of performance and stock price in the first half of the year. Among them, the share price of Wondershare Technology doubled in less than two months. The stock price rose by 299% in the first half of this year, Kunlun Wanwei’s stock price rose by 179% in the first half of the year, and Meitu’s stock price rose by 118% in the first half of the year. HKUST Xunfei’s share price rose by 107% in the first half of the year, and Tom Cat’s stock price rose by 106% in the first half of the year.

In the previously released mid-year report notice, Wondershare Technology expects that the operating income in the first half of the year will increase by more than 30% year-on-year, and the net profit attributable to shareholders of listed companies will reach 35 million to 52 million yuan, a year-on-year increase of 199.23%-344.57%; HKUST Xunfei said that the operation in the second quarter reversed the unfavorable situation of the decline in the performance in the first quarter. It is expected to achieve operating income of about 4.9 billion yuan, an increase of about 10% over the same period last year, and an increase of about 70% from the previous quarter. It is expected to achieve a gross profit of about 1.9 billion yuan. billion, a quarter-on-quarter increase of approximately 60%.

Meitu also expects that the group may obtain a net profit attributable to owners of the company of approximately RMB 220 million to RMB 265 million in the six months ended June 30, 2023, turning losses into profits year-on-year.

In the announcement, Meitu stated that the significant increase in adjusted net profit was mainly attributable to the rapid growth in revenue from the Group's VIP subscription business for images and imaging-related applications driven by AIGC. Relevant analysts said that compared with previous coin speculation, AIGC has allowed Meitu to return to a more reliable main business.

ChatGPT traffic decline, an expected adjustment

The launch of ChatGPT has broken multiple perceptions in the Internet industry. On the one hand, it is the experience of the product itself, and on the other hand, it is the rapid growth of users. It took only two months to reach hundreds of millions of monthly users. In comparison, it took Facebook two years.

From GPT3.5 to GPT4.0, a more complete experience has also raised everyone's expectations for future AI. In the early morning of May 19th, OpenAI released the iPhone version of ChatGPT, which is free and has no advertisements. The Android version has also been released a few days ago. Some users said after the experience that the "iPhone moment" of ChatGPT has come.

Source: Similarweb data

However, just over a month after the release of ChatGPT on the mobile phone, doubts came. Statistics show that in June 2023, ChatGPT's visits fell by 9.7% month-on-month, the first time since its launch. This negative growth also formed a significant contrast with the growth rate of 131.6% in January. It is worth noting that not only ChatGPT, but almost all AI chatbots have encountered bottlenecks in user growth.

In this regard, Professor Shen Yang believes that for early adopters, the degree of information has been reduced, and they will pursue practical and more in-depth interactions. Moreover, the use of ChatGPT actually has a certain threshold. Only by using language to prompt can produce very high-quality content, and users with this ability are limited.

Tang Fangxin, head of strategic development at Wondershare Technology, also gave the same answer, "At the current stage, the threshold for using ChatGPT is very high. Although it can communicate with users, it also requires users to have another logical framework for thinking. If you need to get To get the answers you want, you need certain questioning and communication skills.”

Not only the traffic growth of ChatGPT has peaked, but the AIGC concept stocks mentioned above also suffered a wave of decline in June, among which Wondershare Technology’s single-month decline was nearly 20%. In the face of stock price fluctuations, Tang Fangxin did not care too much. He believes, "Compared with short-term stock price fluctuations, what we need more is to make good products and meet the expectations of users. This is what we pursue. Short-term rises will continue High, if there is no product and no actual technical support, it will be a bubble.”

Application ecology is the key to breaking through, and it will take time to release value

The decline in ChatGPT traffic and the cooling of large models do not represent the decline of AI, but instead herald the coming of a new stage of AIGC. With the support of the large model architecture, more exploration on the application side has become a new topic in the industry.

Tang Fangxin pointed out that the early adopter stage of ChatGPT has passed, but the expectations for the entire industry will not decline, and now it is an opportunity for tool products. For application manufacturers, what they need to do is to help users do training and continue to lower the threshold.

Previously, Wang Xiaochuan, the founder and CEO of Baichuan Intelligence, once said that generative AI will mainly have three major application directions, one is to help the existing industry upgrade through API calls or privatization deployment, the other is intelligent assistants, and the third is AI. is to help create the unreal open world.

IDC predicts that the scale of China's AI market will grow at a compound annual growth rate of 24.4%, and is expected to exceed US$18.43 billion in 2025. Acumen Research and Consulting, a consulting firm, also predicted that the AIGC global market size will reach US$110 billion in 2030.

The upper limit of AIGC is still increasing. It can be seen that many large-scale model manufacturers are connecting with enterprises in various industries for applications, such as smart home, painting, video editing, and live delivery. According to Gartner research, by 2030, 90% of digital content will be generated by AI. It is estimated that by 2032, the global AIGC market size will increase from US$10.8 billion in 2022 to US$118.1 billion, a 10-fold increase in 10 years.

Image source: Tom cat announcement

At present, a number of listed companies have also begun to raise funds for AIGC application research and development. Among them, Tom Cat plans to raise no more than 2.3 billion yuan, which will be used to invest in the Tom Cat family IP series AI interactive terminal product research and development projects. On July 24, it also revealed on the interactive platform that the AI products created by the company's domestic R&D team have initially realized the functions of situational dialogue, oral English enlightenment, AI object recognition, and AI drawing.

Image source: Wondershare Technology Announcement

In June, Wondershare Technology plans to raise a total of no more than 1.101 billion yuan from no more than 35 specific objects to accelerate product and technology upgrades in AIGC-related fields. The official said that its main overseas products have been connected to Azure OpenAI services in the first half of the year, and products such as Wondershare Miaoying and Yitu Graphics have launched AIGC functions.

Image source: Announcement of Insay Group

Also in June, Insai Group also planned to issue no more than 33 million shares and raise no more than 645 million yuan by way of inquiry. After deducting the issuance expenses, the funds raised will be used for marketing AIGC large-scale model research and development and application projects and supplementing working capital.

The technical capabilities of large models are not very perceptive to most C-end users, and the freshness will soon pass. Only by transferring the technology to the application layer and changing the user experience can greater value be generated. For example, the popular AI-generated photo "Miaoya Camera" during this period has made more users feel the revolutionary power of AI. As Shen Yang said, "Users have also discovered that different AI products have different advantages. The application value of large models is still great, but it will take some time to release the value."

However, one thing to note is that AIGC is not just for chatting or drawing. What the market needs are differentiated value applications, not homogeneous products. Of course, if we want all users to feel the universal power of AIGC, we still need to wait for the emergence of a killer application. (This article is the first release of Titanium Media App, author/Du Zhiqiang, editor/Zhong Yi)