Xiao Feng: The Future of AGI Business Organizations

** SOURCE: ** UNICORN ** DIGITALIZATION **

On June 11, 2023, Towards a New Era of General Artificial Intelligence and the new book launch of "General Artificial Intelligence" published by [China Translation Publishing House] was held in Beijing. Xiao Feng, chairman of HashKey Group and Wanxiang Blockchain, and co-chairman of Metaverse 30 Forum, delivered a keynote speech on the topic of "The Future of AGI Commercial Organizations".

The full text of the speech is as follows:

The topic I want to share today is "The Future of AGI Business Organizations". What is the relationship between AGI, Web3 and Metaverse?

The future of AGI business organizations, I think this topic may be discussed less, but it will have a huge impact on the future development path and degree of success of AGI. How do we organize these technologies? How to productize and service it? How can it be applied in the business field? How to even commercialize AGI? All of these are closely related to business organizations.

The origin of this topic is a kind of worry, even fear, about the monopoly of AGI. You will find that AGI may be more concentrated and monopolistic than Internet platforms. At the Beijing Zhiyuan Conference yesterday, Professor Huang Tiejun, the dean of the Institute of Artificial Intelligence, also mentioned in his speech that the world may only need three large models. If there are only three large models in the world, then the degree of monopoly is obviously much higher than that of global Internet platforms, because there are at least a dozen Internet platforms.

How do you view the high concentration and high monopoly that AGI can see in the future?

First of all, we must objectively and rationally admit that the monopoly of AGI is an inevitable result of the laws of the digital economy. I have expressed similar views before. The products and services of the digital economy have a law, that is, high fixed costs, low marginal costs, and even marginal costs approaching zero. If the marginal cost of a product or service approaches zero, it will inevitably lead to high concentration and high monopoly, which is dictated by economic laws. However, in the face of such a law of digital products and services with high fixed costs, low marginal costs, or even zero marginal costs, it does not mean that it will inevitably move towards monopoly or high concentration and high monopoly, nor does it mean that it will inevitably Let some people get huge profits and benefits.

We are very pleased to see that OpenAI has announced a unique equity structure, which involves the distribution of equity. At first, OpenAI was a limited liability company, which can also be called a joint stock company. But it has restrictions on the income of all shareholders and investors, that is, there is an upper limit of income. Over time, OpenAI will transform into a for-profit limited company. When Microsoft's earnings hit $120 billion, Microsoft's stake in OpenAI will cease to exist. OpenAI will become a non-profit organization. Becoming a non-profit organization is only the first step. After becoming a non-profit organization, I see that the ultimate goal of OpenAI is to become a common public infrastructure for mankind, an open source, open, license-free, and trust-free, like the Internet TCP A common human infrastructure such as the /IP protocol.

This structure is a very new structure. Before that, Wall Street would not set up such a structure. Only Silicon Valley, only those technology companies, especially companies like OpenAI that are becoming more and more digital, will adopt this architecture. They will understand their social responsibilities, and they will also understand that when leading OpenAI and GPT to the AGI era, they need to answer people's concerns about monopoly and the possibility of bringing excess monopoly profits to a small number of people-everyone's concern, everyone's suspicion , and even everyone's outrage about this. This is the path we see OpenAI taking—from a limited company, a profit-making limited company to a non-profit organization to a general infrastructure, and finally it will become something similar to the Internet TCP/IP protocol stack. This is a very good development trend.

I think that all AGIs in the future, no matter whether it is a large model or others, will eventually go on this road and complete their mission along such a model. It is unlikely that there will be a monopoly profit like the Internet platform. True monopoly profits.

Looking back at the development of the past few decades, we can see that the form of open source, open, permissionless organization has been changing. Of course, the earliest is the TCP/IP Internet protocol stack. Now we don’t need to pay anyone to use the IP protocol and TCP protocol, and we don’t need to ask anyone for permission or approval. This is the first A form of open source, later the Internet operating system, the Linux Foundation also falls into this category. They are representative of the Internet age.

Later, the second type of open source organizational form appeared, that is, software tools or application protocols. They form some open source foundations. The code of these foundations is open source, but the use may require permission, and value-added services will charge fees, such as consulting fees or service fees. Some open source organizations have even become listed companies and were later acquired at high prices, with a value reaching tens of billions of dollars. This is the second type.

The third category is the protocol of the blockchain in the last 10 years. The protocol of the blockchain is completely open source, open, license-free, and trust-free. Anyone can use it, and anyone can use it. The original agreement can be forked, and anyone can build their own application on this agreement without any approval. However, it is somewhat different from the previous two types of open source organizations, and it has a built-in functional Token. Through this functional Token, the use value of this network is captured, and through the standardization and sharing of usage rights, it becomes a functional Token, and economic incentives are used to distribute benefits. This is the third category.

**In the fourth category, we can see that OpenAI has built a brand-new equity distribution framework and a new model of property rights licensing so far. **It can be seen that the agreement in the future AGI era should gradually be a non-profit, open source, open, license-free, and trust-free organization, such an equity structure or intellectual property management structure. When that step is reached, technologies like the current OpenAI and GPT will become a common infrastructure, just like the TCP/IP protocol and the blockchain protocol, the AGI protocol will also become a common infrastructure.

After that, we can foresee the digital world, the virtual world, and the "mirror world" recently mentioned by Kevin Kelly, or the world of dozens of Metaverse books published by our China Translation Publishing House. For the metaverse, it may build its own digital world based on three common infrastructures, virtual world, mirror world or metaverse. The first common infrastructure is the TCP/IP protocol stack of the Internet; the second common infrastructure is the blockchain protocol stack represented by Ethereum, which is stacked layer by layer from L0, L1, L2, L3, and L4 The protocol of the blockchain; at the same time, the third common infrastructure is the protocol of AGI. These three common infrastructures build our digital world, virtual world, mirror world and metaverse. This is an immature view I have observed myself recently, and I call it "The Future of AGI Business Organizations". This is what AGI business organizations may look like in the future.

Therefore, Web3 is a foundation of the AGI era, the foundation of the digital world, and the foundation of the digital economy. Through Web3 and other decentralized technologies, or through blockchain distributed ledgers and blockchain tokens, we can turn AGI technology into a general infrastructure instead of a monopoly business organization.

Finally, I would like to respond to Yi Huanhuan’s point of view. He estimates that the AGI era may appear in the next two to three years. I try to give a more specific time—that is, 2025. 2025 may be the real singularity or explosive point in the era of AGI general artificial intelligence, because the maturity of the three general-purpose infrastructures mentioned above all point to 2025.

First of all, I think GPT-5 is very likely to be launched in 2025, which is the first point considering the GPT-4 we currently have.

Second, augmented reality (AR) hardware devices represented by Apple will begin mass production in 2024, and by 2025 there may be more than 100 million global users or sales of more than 100 million hardware devices. Globally, having 100 million hardware AI devices will be an inflection point, whether from a user, usage or business perspective.

At the same time, blockchain technology, especially Ethereum's sharding, layering, and grading technologies, will enable its performance to reach 100,000 or hundreds of thousands of TPS per second in 2025. Achieving such performance will support building AGI applications on decentralized networks, decentralized ledgers, and decentralized financial systems. Therefore, the roadmaps for various technological developments all point to 2025, which is a time point worth looking forward to. It is only about two years away from now, and it is very worth looking forward to. And the arrival of the singularity is indeed imminent.

I think I'll share that much, thank you all!