AI Redefines "Me": Everyone Can Become a Scientist After Blending with AI | 36Kr WISE2025 Business King Conference

From November 27th to 28th, the 36Kr WISE 2025 King of Business Conference, known as the "annual technology and business compass", was held at the Conduction Space in the 798 Art District in Beijing.

This year's WISE is no longer a traditional industry summit, but an immersive experience centered around the theme of "technology-driven hit short dramas". From AI reshaping the boundaries of hardware to embodied intelligence opening the door to the real world; from brand globalization in the wave of going global to traditional industries equipping with "cyber prosthetics" - what we present is not only trends but also the insights refined through numerous business practices.

In the following content, we will dissect the real logic behind these "hit dramas" frame by frame and explore the unique "business scenery" of 2025 together.

Dear friends:

The following is the content of the conversation, edited by 36Kr:

Feng Dagang: Well, I'm very glad to invite Mr. Sun. We know that the past year has undoubtedly been a year of explosive growth in AI applications. There are many applications that we can feel, such as various AI games, AI conversations, and AI companionship. However, there are also many that we don't perceive as clearly. I don't think these are unimportant. In fact, they have a significant impact on the entire industry and even on the fate of humanity. As for AI for Science, I'm not sure if there's a clear explanation for it, which refers to AI development for scientific research. I think it's a very attractive field. For example, in Silicon Valley, this field is highly regarded. I know that OpenAI has also developed relevant projects this year. Actually, I'd like to know what Chinese entrepreneurs are doing in the field of AI for Science. Are they as good as, or even better than, their American counterparts? I think the most suitable interviewee for this topic is DP Technology. I heard that the concept of AI for Science was put forward by you or your mentor, and you've been working on it for 7 years.

Sun Weijie: Yes, it was in 2018 when Academician E Weinan, the mentor of Lin Feng and me, first proposed the concept of AI for Science at Peking University. DP Technology is probably the world's first company to systematically layout this field, perhaps without a doubt.

Feng Dagang: Right, the first one to propose it. Mr. Sun, you have a liberal arts background, but you've been the CEO and founder of technology companies for many years. I'd like to know how you view the field of AI for Science from the perspective of the intersection of liberal arts and science. What's its value to us?

Sun Weijie: Well, I'm very grateful to Mr. Feng Dagang for the invitation. It's also a great honor for me to participate in today's dialogue.

Let me share a little-known fact. In the traditional academic classification, what we currently consider as liberal arts and science are actually collectively referred to as liberal arts abroad. So, although I have a liberal arts background, I think the disciplines I've studied have a lot in common with the academic and scientific spirit of many science disciplines at the fundamental level.

In simple terms, AI for Science means using AI to assist human scientific discovery. Its ultimate vision is whether we can create a series of AI scientists based on AI technology, who can make scientific discoveries like human scientists, or develop an intelligent system capable of autonomous scientific discovery. This might be the ultimate goal of AI for Science.

I think it has several extremely important values for the fate of humanity:

Firstly, many people may not have thought about or overlooked the fact that scientific research itself is also a production activity. It conforms to our definition of production activities. We input production factors, go through a complete production process, and obtain our production output, which is scientific achievements.

In the past, almost all production processes have become assembly lines. Why can't scientific discovery be made into an assembly line?

Feng Dagang: Actually, many people think it's normal for production to become an assembly line, but how can scientific research achieve that?

Sun Weijie: That's a very good question. First of all, we recognize that scientific research is also a production activity. As long as it's a production activity, we can turn it into an assembly line. How to do it? The answer to this question exists; it's just that we haven't really thought about it in the past. In the AI era, the first important significance is that scientific research in many important fields can really take the form of an assembly line, and a series of high-value scientific achievements can be produced in batches. For example, if we need a certain type of drug or material, we just tell AI our requirements. As the teacher from JD.com said, I really like the idea that you just need to raise the question, and regardless of how complex the subsequent process is, AI and related robot technologies will solve the problem. So, this has a very important impact on the fate of humanity.

The second important impact on the fate of humanity is that the ultimate questions that the global community of shared future for mankind is concerned about will be better answered. For example, the issue of aging, the problem of infinite energy, the desire for interstellar travel, and the need for faster and more timely communication. In essence, these are what humanity has been pursuing for a long time. Whether it's aging, interstellar travel, or better industrial development, it essentially requires more powerful drugs, more advanced materials, a deeper understanding of life itself, and further exploration of some of the most perplexing problems in physics. With such a scientific research assembly line, we really have the hope of answering these questions well.

Of course, I think there's also a third very important significance for the fate of humanity. When scientific research becomes a production model like an assembly line, the threshold for everyone to participate in scientific research will be greatly lowered. In the past agricultural era, more than 60% of the population was engaged in agricultural production. In the current industrial era, most people are engaged in industrial production. Now, in the information technology era, more and more people are engaged in IP and art-related production, and this is also a greater demand in the future. That is, more and more people will join scientific production.

The threshold for resources, knowledge, and tools in scientific research will be significantly reduced. Correspondingly, the group of people participating in science will increase. Those who have many innovative ideas in science and can put forward very ingenious scientific requirements and engineering architectures will be able to generate more value in scientific research. Of course, further, scientific achievements will also be shared by more people. So, in essence, it will bring about equality in knowledge and science.

I think these three significances are inevitable, at least from the perspective of a liberal arts person like me.

Feng Dagang: As we can see, AI tools are becoming more and more powerful. Many things that we thought were impossible in the past can now be achieved through AI. For example, in AI programming, as long as we put forward our requirements, it can help us write code. I have a metaphor that might help you better understand this, but I'm not sure if it's accurate. We all know the story of Edison. He didn't invent electricity, but he invented the light bulb. When we were kids, we read about how Edison invented the light bulb. He conducted thousands or even tens of thousands of experiments.

Sun Weijie: He tried more than 7,000 materials and conducted nearly 300,000 experiments.

Feng Dagang: It might have taken him many years. So, today, if we tell DP Technology that we want to invent a light bulb and need to try countless materials, and it knows the conductivity and resistivity of all materials, can it help us invent the light bulb in just one month?

Sun Weijie: Yes, it's very likely.

Feng Dagang: So, today, AI can not only make you a good photographer, a good director, or a good game producer but also turn you into a scientist?

Sun Weijie: Yes. For example, for the requirement you just mentioned, Edison relied on his own experience and kept looking for suitable materials. He would basically take any material he saw, twist it into a filament, and test it on the lamp. Now, AI will first integrate and search through all the existing academic achievements globally to see if there are any potential suitable bases. Then, AI will conduct a huge amount of calculations. It might calculate more than 100 different materials in the computer to see if they have the characteristics to become a new type of filament. Of course, in the end, AI might still conduct dozens to hundreds of tests, which might be completely carried out by machines. During this process, feedback will be given to the AI model for further learning.

Feng Dagang: So, I suggest that you use Edison's portrait as your logo when doing consumer-oriented promotion in the future.

Sun Weijie: It's worth considering.

Feng Dagang: In fact, great progress has been made in the field of AI for Science in recent years. We saw that last year's Nobel Prize in Chemistry was awarded to an AI team, the DeepMind team under Google. I think this clearly proves that the scientific community highly recognizes that this can help science make great strides forward. I'd like to know how the AI for Science industry in China is developing?

Sun Weijie: First of all, I think China is at least not lagging behind the United States in AI for Science. In terms of the proposal of this concept and the construction of the entire industry landscape, I think we are more systematic than the United States. The relative disadvantage in China is the lack of large investment in some major and high-profile issues. As a result, in issues like AlphaFold, we don't have as much resources as Google, Microsoft, DeepMind, and NVIDIA, which pour a large amount of money and resources into one problem and can quickly achieve a high-profile result like AlphaFold. Instead, I think we have chosen a more difficult but more correct and more promising path, which is the construction of infrastructure. For example, as we analyzed earlier, there are three basic infrastructures in scientific research, which we simply call: read, calculate, and do. The first is the scientific literature and knowledge base, the second is a series of infrastructures related to scientific computing, and the third is the automated laboratory. We can see that Chinese entrepreneurs, including academic institutions and even many large institutions led by the government, have done very well in the construction of these infrastructures. Of course, the contribution of DP Technology is also indispensable. In the scientific knowledge base, just as AlphaFold solves the problem of proteins, our molecular, atomic, and gene large models are among the best in the world. Not only DP Technology but also many commercial companies in the industry and many large academic institutions have made a lot of arrangements in the field of automated and intelligent laboratories.

Why do I think the construction of infrastructure is a more promising path? It won't create a big splash like AlphaFold. However, after the infrastructure is built, more downstream results will emerge from scientists based on this platform. It's like building a highway and laying a network. We can't predict where an explosive highlight will appear. But such infrastructure is essential for everyone to conduct basic research in the future.

Feng Dagang: We just said that China is at least not lagging behind. Are there any unique advantages in China that can make us lead?

Sun Weijie: I think the first advantage is that currently, 50% of the global AI talents are of Chinese origin, and more than 50% of the papers in basic science are published by Chinese people. So, the large number of talents is the first major advantage. The second is our hardworking, brave, and assiduous work style and high efficiency, which is undoubtedly a basic advantage.

The second advantage is that when it comes to AI for Science in China, it must be deeply integrated with the physical world, whether it's the integration and implementation with industries or the establishment of physical laboratories, the construction of chemical infrastructure, the biological supply chain, or the supply chain of physical laboratories. Only China has such a complete supply chain system.

The third advantage is that if we look at national policies, China's development goal before 2035 is to achieve high - level self - reliance and self - improvement. By 2030, we are also aiming for self - reliance and self - improvement. High - level self - reliance and self - improvement must be led by science and technology, which means we must rely on our own infrastructure to solve our own scientific problems and serve our own scientific and industrial values. This also means that our infrastructure needs to be greatly upgraded. In fact, we were relatively backward in the previous generation of scientific infrastructure, including scientific databases, software, instruments, and even national - level large - scale facilities. To be honest, we are not afraid to admit that we were behind Europe and the United States in the previous generation. However, this actually gives us less historical burden. Just like we skipped physical credit cards and directly entered the era of digital currency, WeChat Pay, and Alipay Pay from paper currency payment.

Feng Dagang: So, in the AI era, China has a greater chance of winning the Nobel Prize in Chemistry and Physics because of AI for Science.

Sun Weijie: Since the Nobel Prize rewards the achievements of the past 20 to 30 years, I think at least after 2035, we really have the possibility of producing a large number of Nobel Prize - winning achievements based on the infrastructure of AI for Science, just like Japan's scientific layout in the past 20 years.

Feng Dagang: Okay, the next question is that many companies regard AI for Science as a daunting task because it's very difficult and requires a huge amount of investment. So, many people think that only large companies like Google are suitable for making such high - stake infrastructure investments. Another view is that even if a startup company develops a good AI for Science product, its achievements will eventually be taken away by large companies. What's your opinion on this? It's quite a big pressure for a startup company.

Sun Weijie: It's also a good question. When it comes to investing in frontier technologies, if you haven't clearly recognized the business opportunities and are just making a layout as a frontier technology, large companies definitely have obvious advantages because they don't have to calculate the input - output ratio.

However, I think we can never ignore the fact that all large companies have grown from small companies. And in any new industry, there will always be new giants emerging in areas that large companies haven't noticed. So, in the field of AI for Science, the core is who can quickly complete the cold start and the iteration of the growth flywheel. This is actually the most crucial point in the entire field. It's not about who has more strength but about who can first iterate better products, attract more users, and gather better data to make the flywheel spin, which will generate better acceleration and potential energy. Because when competing with DP Technology, it's not Google using its entire $3 trillion company but a small AI for Science team within Google.

Feng Dagang: I understand. I'm reminded of an entertainment news story. A female celebrity was being interviewed, and someone asked if she wanted to marry into a wealthy family. She said, "I don't need to marry into a wealthy family. I am a wealthy family." So, I think Mr. Sun's answer is that although we are competing with large companies, who knows, maybe we'll become an even bigger company in the future.

Sun Weijie: Of course, we should have dreams, but we also need to do our current work more steadily.

Feng Dagang: Okay, now let's talk about some more practical issues. Most startup companies fail because of cash - flow problems and lack of commercialization. I'd like to know how you view the commercial value of the AI for Science field?

Sun Weijie: I think we can look at it in two ways: its future huge potential and its current commercial value. There's no doubt about its future huge potential. Since scientific research is the most fundamental driving force for social innovation, if the efficiency of scientific research can be improved several times through AI, it means we can continuously generate new scientific knowledge, new scientific discoveries, and IP, which can be further implemented in our industrial closed - loop.

In reality, most people are a bit far from scientific research. After all, in the long history, the number of people truly involved in scientific research has been very small, perhaps only a few million. The global annual investment in scientific research is about $2.8 trillion, and China's investment is 3.6 trillion RMB, accounting for 2.7% of GDP. What level is this? China's investment in healthcare is about 4%, in education is also about 4%, and in military expenditure is only about 2%. So, we should never ignore that scientific research itself is a market with very high commercial value in reality. It's just that fewer people have participated in this process before.

Correspondingly, it means that the research funds corresponding to each intelligent brain and each scientist are a huge amount. In reality, the four major battlefields, basic infrastructures, or important business models of scientific research are scientific databases, scientific software, scientific instruments, and outsourced R & D services, that is, CRO. If you pay a little attention to these four markets, you'll find that there are very respected business giants in each of these four fields. And I think based on AI for Science, many of these businesses are worth being redone.

Feng Dagang: Yes, there are many AI entrepreneurs here. The thing