Wang Sheng of Inno Fund: The new Combo cycle appears, and the "endgame" of AI is general artificial intelligence

Wind up AIGC

AIGC has become the new business watershed of our time, leading the Cambrian explosion of content and creativity. Whether it is the frontier technology industry or the entire economic and social pattern, it will produce huge changes. In order to allow entrepreneurs to better embrace this era, the dark horse of entrepreneurship will take China's AIGC No. 1 service platform as its own responsibility, and conduct a series of reports from the perspective of software and hardware platforms, investors, industry applications and other dimensions by interviewing a number of well-known entrepreneurs, investors, scholars and entrepreneurial practitioners in the industry, so as to discuss new opportunities in the era of large models and show new industrial strength. This is the first article.

The guest of honor in this issue is Wang Sheng, partner of Inno Angel Fund Beijing. Inno Angel Fund focuses on new energy, advanced manufacturing (semiconductor/robotics/aerospace, etc.), life technology, new generation information technology and other fields, mainly investing in the angel stage and pre-A stage, Inno has invested 70%+ projects in the next round, and many projects have launched IPOs. As a well-known early-stage investment institution in China, it has invested in more than 500 innovative projects and established a three-dimensional entrepreneurial service system starting from Tsinghua alumni.

Wang Sheng is the earliest Internet practitioner, who worked in Yinghaiwei and Sohu, and introduced and formulated China's Internet traffic statistics standards; He initiated and founded a number of Internet companies, which were successfully financed and acquired. In the investment industry, it focuses on AIGC, Metaverse, Web3, and has invested in many well-known projects such as Shenyan Technology, Walker AI, Zhongke Deep Intelligence, Future Black Technology, Tuniu Travel Network, etc., and won many honors such as "2017 Zhongguancun Top Ten Angel Investors", "2016 New Investors", CLPA "Best Post-70s Venture Capitalist" for two consecutive years, "36Kr Discovery Investor in China's New Cultural and Entertainment Field" and so on.

In the live broadcast of "Dark Horse Theory" on May 24, Wang Sheng gave an in-depth discussion on the theme of "Venture Capital Opportunities for a New Generation of AI", discussing issues such as large models, domestic substitution and AI entrepreneurship methodologies.

The following is the transcript of this live broadcast:

01

The four stages leading to AGI

The emergence of ChatGPT has ushered in a new outlet for China's venture capital circle. Due to the earlier observation of this wave of AI, Inno Fund has become the most active investment institution in this field, investing in many high-quality companies such as AI and Deep Space Symphony, and has deep insights into the new generation of AI venture capital opportunities including AIGC.

In the view of Wang Sheng, partner of Inno Angel Fund, the end game of AI is a problem that investors need to pay attention to. "Every time I enter the industrial industry to invest, I first want to see what its end game looks like", he believes that "starting with the end" can better speculate and judge the development rhythm and path of the industry. ”

To this question, Wang Sheng's answer is "general artificial intelligence", which he believes will become the common goal of academia, industry and investment, and divides the "endgame road" into four stages.

The first is the AI 1.0 phase, which refers to the decade from 2012 to 2022. This stage contains two larger paradigms, the first is to treat artificial intelligence as a "child", constantly forming a system of knowledge and instilling it into artificial intelligence, in this paradigm, AI is more like an expert system, which is also a unified artificial intelligence development paradigm before 2012; The second paradigm is to give enough data to artificial intelligence to let it learn on its own, but it fails to succeed due to insufficient computing power and insufficient data.

"AI 1.0 is like an animal, not yet creative, just aware of the world and the environment." Wang Sheng mentioned in his speech.

The second is the AI 2.0 stage, which is the initial AGI opened by chatGPT, which is also an important stage we are currently in. Wang calls this stage "Human Brain in Digital World," which, like the human brain, can create, produce, and even change the world. Great companies of trillions or even $10 trillion will also be produced at this stage.

The third is the AI 3.0 stage, Embodied AI, which is a robot with both strength and intelligence. Once it enters the era of embodied intelligence, the efficiency of AI will far exceed that of humans, and human beings will think more deeply and explore the meaning and value of their own existence.

Finally, there is the era of AI 4.0, that is, general artificial intelligence, which will have AGI (AI Being) with free will. For the first time, humans will think deeply from the perspective of the creator, such as whether AI Being is a slave, tool, partner, enemy, or master of human beings, and Alignment (alignment of values).

02

Why is AGI fascinated by VCs?

Why is this wave of AI fascinating VCs? Wang Sheng believes that AGI is a new generation of GPTs (general purpose technologies), that is, general purpose technologies.

At present, economists have summarized a total of 24 general-purpose technologies in human history, which are both universal and widely used, and will also give rise to a large number of innovations. The most important general-purpose technology recognized in human history is steam first, electricity second, and Internet information technology.

Why is universal technology important? Because it has great economic significance – it can interrupt and accelerate the normal process of economic growth. This brings us to the economist Schumpeter's theory of "creative destruction", which means that everything in the past will be broken, bring about the restructuring of industrial technology, and promote nonlinear economic growth.

The essence of creative destruction is a technological paradigm revolution, and this huge change will start a new "Combo cycle". The Combo cycle is a long cycle of 50-60 years in the developed commodity economy found by the Russian economist Kondratiev after analyzing a large number of statistics of the British, French, American, German and world economies. Wang Sheng believes that general-purpose technologies will promote technological paradigm innovation, and technological innovation will bring about a new Combo cycle.

Looking back, there have been 6 Combo cycles in human history. The first two were dominated by the British, and the Americans began to dominate technological development in the third Compo cycle, "especially after World War II, when the fourth and fifth waves were firmly controlled by the Americans." Wang Sheng believes that the Combo cycle has shaped big countries, "As China, we hope to become a country with the greatest value to mankind, and hope to lead human innovation and guide human innovation and development." Then we must seize the Combo cycle, in these decades, we can run to the front and better promote the development of human society. ”

03

Venture capital opportunities for AI 2.0

In this wave of venture capital opportunities, the keywords given by Wang Sheng are "multimodal or cross-modal large model", "vertical domain large model", "AI infra infrastructure", "AI old scene transformation", "AI to create new scenes".

From the perspective of AI infra infrastructure, infrastructure that needs higher energy efficiency in the future will also produce many attempts at new paradigms, including "storage and computing integration", "edge-side inference", and "chiplet" multi-core technology.

In the field of large models, investors will pay more attention to the direction of language large models, multimodal, and vertical knowledge domains. At present, the CoT thinking chain and ToM mind emerging in the language big model have attracted much attention, and many scientists believe that GPT has the ability to reason logically. In the field of language big models, Inno invested in Shenyan Technology, which was jointly incubated by the Natural Language Processing Laboratory (THUNLP) of Tsinghua University and the Beijing Zhiyuan Artificial Intelligence Research Institute (BAAI) to create a new generation of intelligent text information processing engine based on large-scale pre-trained models.

In the direction of multimodal large models, Wang Sheng is more optimistic about the fields of image, video and 3D, he believes that the investment and entrepreneurship time in these subdivisions is more appropriate, and the volume of the industry is sufficient; In the direction of the vertical knowledge domain, AI4Material, AI4Energy and other technologies also have certain representatives.

In addition, the CV large model of multi-modal large model is also developing rapidly, and can be applied to multiple scenarios such as autonomous driving, medical treatment, industrial vision, and robotics in the future. In this subdivision direction, there are currently representative enterprises such as WeRide Zhixing, Zhixing, inferVision, and Aqiu Technology. In terms of image generation, Wang Sheng believes that the biggest challenges are the unstable paradigm and the realization of CV and generative integration technologies. "There will be a lot of change here, and change means entrepreneurial opportunities, but the challenge for the average entrepreneur is very big." It emphasized.

In addition, there are two other directions worth paying attention to. First, AI transforms old scenarios, there are many opportunities for AI to replace labor, but these opportunities are more inclined to industrial giants; The other is that AI creates new scenarios, and the research and development of different enterprises continues to bring new progress, including the "Smaville" town built by Stanford and Google AI Lab as an example, which is automatically driven by AI to form a MINI version of "Westworld"; and Walker AI, invested by Inno, as an AIGC game productivity platform, is also committed to creating new scenarios.

04

Entrepreneurial methodology for AI 2.0

From the perspective of entrepreneurial methodology, Wang Sheng's view is to "let the bullets fly for a while".

Why fly for a while? "Because we have an innovation cycle", this cycle from technology to model to content, we need to wait until the scientific research paradigm reaches the bottom breakthrough, and reconstruct the underlying infrastructure. "For example, the iPhone moment first created tools, and then entered the innovation of business models; When the business model innovation is completed, it will start to roll content and short videos; Finally, a new wave of scientific research breakthroughs begins, and we are now experiencing a new round of innovation cycle, so there is no need to worry. It explains.

Talking about the characteristics of entrepreneurship in the AI 2.0 era, Wang Sheng believes that the main reason is technology equality, entrepreneurs will focus on customer needs, sales and service, and the entrepreneurial effect has been greatly improved. Taking the PC Internet startup era as an example, entrepreneurs need to buy their own servers, build IDC rooms, install operating systems, etc., and often can only use 20%-30% of their energy to solve customer needs. In the era of cloud computing, entrepreneurs only need to focus on service architecture, data design, algorithm application, and can spend 50% of their energy on customers.

Next, once entering the era of AI big models, enterprises only need to pay attention to user needs, "This is very beneficial to society, our waste of funds is very low, and there is no need to repeatedly build wheels." But for entrepreneurs and developers, "think about where our capabilities are, what we should do, and what we shouldn't do." Wang Sheng said.

Dark Horse of Entrepreneurship: What new industries will be spawned by the emergence of big models? Industry opportunities that didn't exist before birth?

Wang Sheng: When the iPhone comes out, no one knows, and eventually a pinduoduo will come out, and no one knows that it will eventually come out with a Douyin. This kind of innovation emergence is not defined by investors, but hopes that everyone will participate, including the government, investors, scientific research community, industry, and entrepreneurs, to give this wave of AI a better entrepreneurial environment. In the process of a large number of entrepreneurs, we can really see new industrial opportunities, and what we can predict is the transformation of old industries, improve efficiency, and give higher productivity.

Dark Horse of Entrepreneurship: In this wave of big model change, what do you think will be the first industry that may be restructured or transformed? What changes should entrepreneurs in this industry make?

Wang Sheng: A lot of changes have already occurred. This wave of AI is currently promoting the two most direct transformations, the first is the understanding and processing of words, so a large number of text workers, self-media, video, screenwriting and creation industries are using this technology; The other thing is the image, which has a great impact on design.

We invest in games a lot, we know more about game companies, and there are really a large number of original artists and designers who have begun to lose their jobs or are on the way to unemployment. It should be noted that it is not AI that eliminates you, at least not yet, and it is the wave of people who will use AI that currently eliminate you, so everyone must embrace this wave of change, and AI is your tool. In the past, you used to hire four or five people to start a studio, but now you may need four or five AI employees, someone to draw, someone to write a script, someone to edit for you, someone to do tasks for you to publish. AI empowers us to become superindividuals.

Dark horse of entrepreneurship: What opportunities will SMEs have based on current innovation?

Wang Sheng: For the vast majority of enterprises, everyone is doing business, and it is OK if you can make money. For example, short video studio live broadcast goods, there are many so-called virtual people, digital people, he automatically help bring goods, help shoot short videos, your cost is greatly optimized.

Why do you have to make products that compete with big manufacturers? Why can't we make good use of the tools provided by the big manufacturers? I think people need to think differently, they don't have to do it, just serve your customers. For example, restaurants have increased their income by 20 or 30% because of the takeaway business, does every restaurant go to a platform like Meituan to recruit riders by itself? Just think about AI from this perspective.

Dark Horse of Startup: From a VC perspective, what stage do you think AI investment, including AIGC, is in? What are the projects that will actually be invested?

Wang Sheng: The big model is a very luxurious team of scientists or very good entrepreneurs and large factories, only those who have the ability to do it are worthy of the trust of VCs, and they are not something that ordinary entrepreneurial enterprises can do. I don't recommend you to think about these issues, these are huge opportunities, but they have nothing to do with you.

We will focus on the creation of new scenes, this will take time, everyone must be patient; As well as the transformation of the old scene, it creates very great value, but often gives the opportunity to the veteran players or giants in the old scene, and it is difficult for entrepreneurs to have a chance in it.

Everyone still has to return to the fundamentals, from the perspective of their own enterprises, not necessarily to change the industry in which they are located, but more to embrace AI, and use AI empowerment to reduce costs and increase efficiency.