With the rapid development of the artificial intelligence industry and the growth of big data technology, the demand for computing power is also growing rapidly. Henan is the transportation hub of the Central Plains, a historical and cultural fortress, and assumes the function of an information hub in the Central Plains.
In 2019, the company set up its second global headquarters in Zhengzhou, Henan Province, and successively built the "Global Digital Gene Bank" and "Intelligent Computing Center" to help Henan build a data computing power highlandIn August 2023, the "big model" of Henan's artificial intelligence industry will be implemented in Henan, providing an innovative foundation for the transformation of Henan's "thousands of industries".
Li Tao, Chairman and CEO of APUS, was invited to attend the "Fifth Advanced Seminar for Private Entrepreneurs in the Central Plains" held by the School of Economics and Management of Tsinghua University on December 2, with the theme of artificial intelligence industry application and value creation.
In this lecture, Li Tao conducted an in-depth analysis of the evolution of artificial intelligence and artificial intelligence large model technology, as well as the opportunities and risks of the development of the artificial intelligence industry in the world today, and gave a detailed explanation of the application of large models, the industrial value of large models and future smart governance.
We first need to understand that AI and digital technologies are different.
When many traditional industries are transforming, they often regard artificial intelligence, digital economy, and digital technology as the same concept, but in fact, the two are fundamentally different. Artificial intelligence is a kind of technical means that simulates human wisdom and thoughts, processes information, solves problems, and achieves automatic decision-making and intelligent control. The so-called digital economy is an economic form based on digital technology, which can help society and enterprises achieve digital transformation, promote the development of enterprises, and promote social progress.
However, there is indeed a certain gap between China and the world's leading level in the development of artificial intelligence, both in chip research and development and algorithms, which lag behind other countries;Coupled with the insufficient number of chips, the "acceleration" of development inevitably slowed down.
If you can't even compare to others in computing power, then how can you keep evolving your AI model with less than others?In terms of APUS, we use an incremental pre-training approach to perform real-time, rapid updates and training on long texts on data.
Actually, APUS does the same.
As a global company, APUS also faces many challenges from international barriers and cultural shocks. From 2018 to 2022, more than 200 products around the world have adopted AI technology, but the actual effect is not ideal. Why?
For the past 4-5 years, APUS has been using smaller models. There is no scale effect of small models, replication effect, and the marginal cost of using small models for application is increasing. Algorithms play a very small role in establishing business processes. In different industries and in different businesses, building a business application with a small model is almost a "starting over".
In contrast, large models have a higher one-time fixed fee;However, with the popularity of various industries and scenarios, the marginal cost gradually approaches zero, and at this time, the value of large-scale models will be reflected.
In November last year, Google released Chat GPT, followed by APUS, and in December, Apple decided to focus all its technology on artificial general intelligence. In April 2023, the entire company has invested in the field of artificial intelligence and launched a 100-billion-level multi-mode large model independently developed in China - APUS. 
In the first half of the year, when talking about artificial intelligence, everyone was talking about large models and computing power, and in the second half of the year, everyone was talking about the application of artificial intelligence.
I think the greatest value of big data is to use artificial intelligence to help the entire industry think and calculate globally.
In the past, people made decisions and lacked the ability to think holistically, nor did they have the ability to calculate globally. As far as agriculture is concerned, it is more difficult to judge the global sowing amount and fertilizer amount in the early stage of production, and the input-output ratio should be determined according to the growth status, harvest and transportation of crops, as well as the transactions in the market. At the same time, there are also certain difficulties in the calculation of variable factors that affect production such as future meteorology. But now, with artificial intelligence, everything is much simpler.
For example, in the popular "autonomous driving", artificial intelligence is still just a "small model". If it is possible to add a large model on the basis of unmanned driving, it will be possible to solve the problem of unmanned driving.
The "big" model needs a lot of data to be "fed" so that it can learn Xi, and at the same time, it can quickly calculate some new things, such as intelligent driving, such as intelligent assistants, and so on. Over time, we will find that many traditional industries, such as those that are difficult to digitize, need the help of AI to play a greater value.
It is widely believed in the industry that the "big model" is the foundation of the Fourth Industrial Revolution. I also often describe "large models" as the "underlying operating system" of artificial intelligence.
However, with the continuous maturity of the big data model, especially the recent release of open AI on November 7, many practitioners are aware of the potential risks of the AI ecosystem.
Let's first take a look at the content of OpenAI's update, and how we can see the challenges and risks faced by the AI ecosystem through this update.
There are three things that have been released at the core: the first is the release of GPT Turbo;The second is the GPT Store, which unifies all GPT entrancesThe third is GPT.
GPT Turbo, essentially OpenAI, provides an open platform for artificial intelligence around the world. This may seem like a good thing from the perspective of AI developers, but many people forget that it is entirely possible for the world's largest model to build a "one-of-a-kind" AI ecosystem.
The release of the GPT store is essentially to build an ecosystem where more people can **, pay, and use the developer's products and technologies. But this time, the GPT store is very different from the past, it is not just a store, but an integrated entrance for all users. Most of the developer's applications are plug-in, integrated with GPT, and will not affect the developer's personal image. Compared with the Apple Store and Google Play, the GPT Store may have a higher commission for users.
GPT is actually a **, it is like a digital assistant, a digital replica, which can solve your problems secretly according to your requirements. But through **, OpenAI can even easily obtain various privacy and confidential data of developers and users, and realize the "three monopolies" of technology, traffic, and data, so as to achieve the business goal of the "AI empire".
As a Chinese AI practitioner, I also expressed my views on intelligent risk and governance issues.
First, economic powers should have their own general AI model base. For example, the United States, China, and even large economies like India and the European Union should have their own general-purpose foundation.
Second, each country, and even each independent civilization, must have a set of AI frameworks that conform to its own national cultural values, and use this framework to accommodate and constrain this huge model.
In addition, he also needs to build a healthy AI industry ecology to provide enough living space for each link in the AI industry chain.
In the early stage of the development of the large model, APUS set the goal of "tailoring a large model of artificial intelligence for China". After experiencing the tempering and practice of the market, the large model of APUS has also clarified its own strategic direction of "growth": adhere to wisdom for good, use the red framework to constrain the large model, align the value, and realize the docking of the application of the large model and value creation.
The development of science and technology requires all links to jointly build and correct deviations in the wave. We should adhere to the principle of "intelligence for good", build a binding value framework on the basis of artificial intelligence, and guide the artificial intelligence industry for good, so as to better serve the production of enterprises, people's lives, and social and economic development, and better create the application value of AI.