Editor's note: In this year's Two Sessions, "AI+" has become a buzzword. How can AI empower the blockchain sector? Zhang Chen, Assistant Professor of the Department of Computing, The Hong Kong Polytechnic University said that since all transaction data on the blockchain is open and transparent, it provides a wide range of application scenarios for AI, and AI technology can learn and analyze massive amounts of financial data, and then realize the transformation of "from data to knowledge" to help investors better understand different blockchain projects, and provide investment guidance and assistance.
【Anchor】Hello everyone, welcome to DDN Business Insider. I am Yunfei Zhang.
The annual "Two Sessions" (National People's Congress and the Chinese People's Political Consultative Conference, "NPC & CPPCC") concluded last week. This year's government work report prioritized technology-related themes, notably the "new quality productivity" concept and the "AI+" initiative. In the realm of innovation and technology, blockchain and digital currency continue to garner substantial attention.
What is the relationship between AI and blockchain-based digital currencies? How can they open new investment opportunities? Today, our guest host, Ding Zhaofei, Chief Analyst at HashKey Group, will share insights with Dr. Zhang Chen, Director of the Artificial Intelligence and AI Laboratory at the Hong Kong Polytechnic University. Now, Let's welcome Mr. Ding.
【Anchor】Hello everyone!
In recent years, numerous companies and organizations have been actively exploring both AI and blockchain technologies. What is the intrinsic relationship between AI and blockchain-based digital currencies? Today, we are privileged to have Professor Zhang Cheng, Director of the Artificial Intelligence and AI Laboratory at the Hong Kong Polytechnic University, to provide us with an in-depth analysis. Welcome, Professor Zhang!
Thank you. It's my pleasure to participate in this discussion.
【Anchor】With the emergence of DICE (Decentralized Internet of Things Ecosystem) and its integration with AR and blockchain technologies, many are curious about the potential innovations and features this convergence might bring. Could you explain, using some practical examples, how AI and digital transformation logic interact with cryptocurrency?
This is indeed an excellent topic. Many of my PhD students share a keen interest in cryptocurrency and the future development of these converging industries. Their research is related to artificial intelligence, and they are publishing papers and looking for research topics. They also come to me with the same question: How should these two be combined?
Yes, primarily, I'm looking at the origins of these two concepts—artificial intelligence and Kratos, which refers to cryptocurrency.
From an academic perspective, AI and cryptocurrency do not inherently share a strong connection. AI probably originated from the early Turing Test conducted on children, which certifies an AI, right? It was during World War II, and then it gradually evolved into the era of big data, what we call the big data age, and then into deep learning, and now the hottest thing is the large language models (LLM), right? That's the developmental process. As for blockchain, including Bitcoin, it was born in 2009, right? Following the 2008 financial crisis, blockchain technology emerged and subsequently flourished. Up to today, it has created a very large entity that we call capital, right?
Yes, but if you look at it from a technical perspective, there isn't an inherent bond between the two. However, when we discuss their interaction, I see two key points.
The first point is that all transaction fees and financial activities on the blockchain are public. For example, my address, when I send you any cryptocurrency, or you send me a token, etc. For the first time in history, financial data – traditionally deemed highly sensitive - is now publicly accessible. Traditionally, banks have heavily protected customer privacy. However, this is the first time we can access such data on a large scale, creating an excellent opportunity for various AI applications.
I believe the most significant applications are in investment fields, where we can analyze this data to derive interesting or useful insights.
Yes, exactly. This is also what we mentioned earlier, as we often refer to in research: the concept of "from data to knowledge." We need data first, right? There is a lot of data available; for example, in this era of external data, there is quite a bit of publicly accessible information. However, financial data is something we can now access relatively easily in the age of big data. I think this is a great intersection point!
Another point of convergence is that with the rise of AI and the gradual reduction of its costs, especially the remarkable aspect is that tasks which previously required hundreds of millions of dollars in investment from large companies can now be accomplished at a much lower cost. This significantly lowers the cost of data analysis, allowing even ordinary investors or non-technical individuals to better understand the data and transform it into knowledge. I believe these two aspects highlight interesting convergence points, both in research and in practical industrial applications.
【Anchor】Yes, you also mentioned that the data on the blockchain...
Centralized patents and centralized data tend to be more efficient. Blockchain emphasizes decentralization to maintain security, stability, and fairness. So, while these two concepts might seem contradictory at a fundamental level, is there room for complementary space?
When discussing decentralization, the inherent characteristic of blockchain is its decentralized nature, right? Many applications utilize their immutability and fairness in the financial sector. However, in finance, I believe this provides an alternative application. Previously, financial transactions required direct connections to banks. For instance, if someone needed to make a payment or perform any financial action, they had to connect with a bank. But now, they can use blockchain technology to connect AI services in a business context. I think this is also a very interesting application.
【Anchor】Perhaps you mentioned earlier that the broader financial sector has always been an important breakthrough area for blockchain. What new projects are we looking forward to seeing implemented?
If I were a project developer, I would likely lean towards using AI technology as a communication tool. For instance, a typical application in fintech is the use of customer service chatbots. However, as a project developer, my ultimate goal would be marketing applications, which means I need to communicate with a large number of users—potential investors, right? The costs of traditional technologies for this purpose are often very high, especially in terms of promotion and management. With the help of large models now, these costs can significantly decrease. Additionally, from the investors' perspective, they can use models to better understand projects.
I can give a simple example that we briefly discussed earlier: in the early days of Bitcoin, all Bitcoin data was public, right? Yes. We could see all these transaction behaviors. At that time, there was a very intuitive and effective investment strategy—monitoring addresses with relatively high holdings. We could identify which Bitcoin addresses belonged to exchanges, right? Yes, and the remaining addresses we would consider as personal investors. Sometimes, we observe a phenomenon where large amounts of assets flowed from individual investors to exchanges. Selling indicates a bearish outlook, right?
Conversely, if large amounts of assets flow from exchange addresses to personal addresses, it suggests accumulation, which could be bullish information. This is an example of how we transform data into knowledge. Data itself is inert—it merely records a transfer from A to B. However, using AI or data mining techniques, we can quantify this information, transforming it into actionable knowledge. This knowledge can effectively help us make investment decisions.
Can AI uncover new knowledge that we do not know? Of course, it can, right? But what are its underlying mechanisms? Actually, we don't know. It may completely depend on how you communicate with it. There is a field called "near," which relates to how I ask my questions. This is something that can turn into a very rewarding task.
Because we don't fully understand large models, we also don't know how great their potential is. That's true. Ihave been concerned that if the same large model generates identical investment conclusions, will this lead to uniform trading actions across the market? It doesn't seem so. Each person cultivating their model will yield different results.
Different investors and institutions might pre-train their own designs. If everyone's data isn't shared, right? Then surely their behaviors will also differ, or their own investment experiences can be instilled in themselves, right? Yes, it will be different.
【Anchor】At this new juncture with DeepSeek, are there any better directions we can explore in either the cryptocurrency or stock market fields?
From an investment perspective, I think about how to what extent I can outperform the market.
【Anchor】Yes, we all know AI will become more powerful, right? That's how it is now. But how about other market participants? If they gradually recognize this potential, should they adopt similar strategies? Or, if you have confidence that your understanding is more advanced or more precise than the market, where is our advantage? I generally reflect on this, especially when making very detailed, short to mid-term investments, as I think this aspect is even more important, right?
Actually, a friend of mine mentioned that when OpenAI was just publicly launched and we could start using it, the first batch of users were mostly computer scientists—early adopters who were into computing. They found this technology to be incredible, as it transcended our traditional understanding, right? My friend, who is very smart, immediately researched OpenAI's equity structure and discovered that Microsoft had invested in them. So, they bought Microsoft stock and quickly made some money. Thisexemplifies an insight that outpaces general market awareness, as not many people knew about OpenAI at that time.
Yes, that's right. Clearly, not many people were aware, and recognizing this technology as disruptive could lead to a revaluation of Chinese assets.
I believe this has indeed sparked a surge in mid-tier development, ultimately leading to a reassessment of U.S. stocks, which subsequently caused a decline in those assets in China as well. This is something we are quite pleased to see—a good outcome, right? Through this point or other breakout products, we can uncover their underlying equity structures or explore different ecological niches and product variations for research and investment.
So, to what extent can we be confident that our understanding will surpass the market? This is actually a difficult question, a low-probability event. However, I believe AI can help us a lot in finding more opportunities. It allows us to recognize that you can ask about anything, which is a good thing.
Currently, I think AI is still in its early stages. Knowledge gained on paper is superficial; true understanding comes from practical application. We need to use AI to invest and see if we can uncover some of the investment opportunities that Professor Zhang mentioned, which could help us get ahead.
【Anchor】Actually, we are also very interested in the applications of AI in areas like anti-money laundering (AML) and regulatory compliance. For example, we know that Tether (USDT) can often be used for unsafe or non-compliant payments. What role can AI play in this context?
From a research perspective, there are indeed many related studies in academia, right? The basic issue is, for example, if I have a payment network with many addresses involved in financial transactions, and once I know some addresses that are problematic, I can gather evidence to prove they are associated with issues. Then, using AI methods or data mining techniques, we can analyze these addresses since they have certain transactional behaviors that align with patterns of money laundering or illegal transactions. We can label these addresses as high-risk.
As far as I know, some security companies currently provide such services. For instance, if you give them an address, they can analyze its safety coefficient based on the addresses it has transacted with. There are indeed products available for this.
However, up to now, there hasn't been a technology that can effectively prevent such issues.
In some cases, even when the evidence is quite conclusive, it can involve companies like Tensor, as mentioned earlier with USDT, which may be used in illegal activities. These companies can assist law enforcement in freezing relevant addresses, but for such actions, you need very concrete evidence. Unfortunately, acquiring this evidence is relatively difficult, as I understand.
Yes, that's correct. AI can serve as an early warning system.
Thank you, Professor Zhang!
【Anchor】OK, thank you. That's all for this episode. Remember to follow us on YouTube or download our APP. I'm Yunfei Zhang, thanks for watching, and see you next time.
Anchor: Laura Cheung | Edited: Kelly Yang, Laura Cheung | Translate: Kato Ip | Proofread: Chris Liu
Related News:
DDN Business Insider | Outlook of Two Sessions: What expectations for new issues & themes are there?
Comment