Dialogue with KIP Protocol Founder: The New Trend of AI Concepts, How KIP Empowers Decentralized AI

Under the wave of digitization and decentralization, blockchain and artificial intelligence technologies are rapidly merging, redefining the way various industries operate. The KIP Protocol team is at the forefront of this technological transformation. By addressing core issues such as on-chain and off-chain data interoperability, revenue realization, and digital property rights protection, it aims to create a non-monopolistic, independent, open, and fair AI ecosystem.

ChainCatcher invited Julian Peh, CEO and co-founder of KIP Protocol, as a guest for the Space to discuss the "New Windfall of AI Concepts."

During this event, Julian Peh delved into the increasingly severe data security issues in the AI field, KIP Protocol's unique philosophy and innovations, as well as future incentive plans. He will also reveal the KIP Protocol team's unique insights into the education industry and the development of AI technology, presenting readers with a forward-looking and inspiring technological exploration journey.

The following is the text content from ChainCatcher Space:

1. Hello Julian, could you please introduce yourself?

Hello everyone, I am Julian, the CEO and co-founder of KIP Protocol. Regarding my background, I can say that I am an experienced internet person. I started my first business in 1998 and have been involved in the Web1 internet since then. From 1998 to 2016, I founded and operated a series of internet companies. Around 2015 and 2016, I sold all my companies, which were basically part of the first wave of internet enterprises, mainly related to media.

After that, I entered the venture capital (VC) and private equity (PE) industry, focusing on investments in Web3, robotics, and AI. In fact, KIP Protocol originated from an AI consulting company that mainly engaged in big data analysis consulting. While consulting clients on AI projects, we discovered some market opportunities and gaps.

Moreover, we found that there were some very urgent issues in the market that could be said to threaten the common interests of all humanity. This is also why we are promoting the decentralization of AI. KIP Protocol was essentially created to address the current problems that we believe must be solved.

2. It is well known that KIP Protocol is the first decentralized Web3 underlying protocol focused on AI, concentrating on data and property rights protection in the AI field. How does KIP Protocol understand the current "Crypto + AI" environment and the increasingly severe data security issues in the AI field?

Because we see the threats currently facing the market, we want to create an organization, not just a company. We hope to raise awareness about the current dangerous situation through this organization. Basically, every time a large AI company (like OpenAI) releases a new model, it always sparks a wave of interest and widespread attention. People are often amazed by the power and innovation of these new technologies. However, these large models are essentially trained using our collective knowledge.

This was not widely known a few years ago, but now more and more people are starting to sue companies like OpenAI and Google, and public awareness is gradually increasing. If you published any content on the internet before 2022, that content may now be included in the training data for GPT-4. This means that our collective knowledge and personal knowledge have been privatized and used to train these proprietary models. These models are then commercialized in various ways, such as charging a monthly subscription fee of $20.

Setting aside this minor unfairness, the greatest danger comes from the widespread application of these large models. Large enterprises will inevitably apply these models to all available scenarios, and as the scale of the models increases, their training and inference speed and capabilities continue to improve. As a result, it will become increasingly difficult for small companies to keep up with this rapid development. Although the entire AI industry is still in its early stages, monopolistic forces have already begun to emerge.

We believe that the core of AI lies in data. Without data, AI cannot continue to progress or demonstrate its astonishing innovative results. The reason large enterprises have reached their current level is due to what we call the "great data knowledge plunder." Before the AI era, most people did not realize that their data had immense technical and economic value for AI.

Therefore, it is essential to recognize that our data should be granted ownership. We should hold ownership of our data, clearly stating that this data belongs to us and taking protective measures. If others wish to use this data, it is not entirely unfeasible, but it cannot be done through theft or plunder.

One of the cores of KIP Protocol is to focus on data. We are committed to serving three parties, which I will explain in detail later, one of which is the data party, i.e., the data owners. So, who are the data owners? In fact, it is all of us. As long as you have any activity on the internet, you are a data owner because you are creating new data every day. Whether it is data generated through daily internet activities or high-value data produced by experts in a certain field writing articles or conducting research, these things that can create economic value in AI application scenarios should be protected.

To achieve this protection, we utilize decentralized Web3 technology. The key to Web3 is ownership. We hope to handle data with the same mindset, ensuring that data ownership is reflected. KIP Protocol provides a series of tools and contracts, as well as an interactive system, to connect these AI assets and ultimately create economic value in the AI field. By deploying data on Web3, we can ensure data ownership and protection, maximizing its economic benefits in AI applications.

3. We see in KIP Protocol's white paper that KIP aims to solve on-chain and off-chain data interoperability, revenue realization, and digital property rights protection, which are also key concerns for users. Could you elaborate on the principles behind these issues and what KIP can bring to users?

The three core elements that KIP Protocol focuses on are models, apps, and data. For example, suppose you have a university textbook that has intellectual property (IP) and is an authoritative textbook. We can place this textbook into a knowledge base, which is one part. Then, a model processes this knowledge, which is the second part. The third part is an app that can help students answer questions or generate exams using this textbook. By connecting the knowledge base of the textbook, the model, and the app, a complete AI product is formed. In a non-monopolistic, independent, open, and fair competitive environment, companies that own the knowledge base, develop models, and create apps are usually different entities because these tasks require different expertise and resources. So, how can we enable these three different industry companies or individuals to collaborate to create economic value?

First, we need to connect the three parties. After connecting, they can all share a portion of the revenue from user payments. Users obtain answers through using this AI product, and the revenue for the three parties comes from user payments. Therefore, the first problem we need to solve is deploying these AI assets on Web3 so that each party can retain its ownership.

Once deployed on Web3, some technical issues arise. The first technical issue is how to connect the data. We use EC3525 format SFTs (Semi-Fungible Tokens) to represent and protect these AI assets. These assets need to be connected, able to interact, and generate answers. At the same time, we need to address the revenue distribution issue.

Thus, we have two problems to solve: the interaction problem and the revenue problem. The revenue problem can be further divided into three sub-issues.

How to record the interactions between these AI assets: For example, when a user asks a question through the app, the app passes the question to the model. The model reads and understands the question, extracts the necessary data from the external knowledge base, generates an answer, and returns it to the app, which then provides the answer to the user. This is a complete interaction process, and each step needs to be recorded.

How to settle accounts: After recording the interactions, the contributions of each party must be settled. For example, in the interaction mentioned above, the contributions of the app, model, and knowledge base need to be recorded in detail and settled to determine their respective shares of the revenue.

How to distribute the revenue: After recording and settling, the revenue is distributed to the model designers, data owners, and app developers. In this way, all participants can fairly share the fees paid by users.

Additionally, there is an important issue regarding ownership and security. Web3 technology itself addresses the issue of ownership, but security still needs attention. The needs of our clients vary; some have very high data security requirements because they handle private data, intellectual property data, or confidential data, while others have relatively low data security needs because they deal with public data.

KIP Protocol can ensure that users' knowledge bases can freely interact with models and apps. After the interaction, these processes will be recorded in detail, followed by settlement, and ultimately revenue extraction. Our approach is to solve the problems of interaction, settlement, and revenue distribution through this complete framework, ensuring that data ownership and security are guaranteed. In this way, our platform can enable different companies or individuals to collaborate and create economic value in a fair and open environment.

4. In the product design of KIP Protocol, how do you balance the distribution of interests among AI app developers, data owners, and AI model designers? How can you better incentivize more AI value creators to participate in your ecosystem? Do you have a similar role as validators to supervise and constrain them?

To answer how interests are distributed, to ensure fairness, each independent asset should have its own pricing. For example, if I am a user with intellectual property (IP) and I have a textbook, then in AI applications, each answer should have a price. This model is called "pay-per-answer," meaning each answer has a cost.

Specifically, the core payment model for AI is pay-per-answer, as the operation of the model consumes resources such as GPUs, so the generation of each answer involves resource consumption. Within this pay-per-answer framework, whether you are a data provider, model owner, or app developer, the core is to conduct business operations within this framework, and the price of each answer needs to be clearly indicated.

To illustrate with a simple example:

• The IP owner of the textbook sets the price for each answer at $1.
• The model owner also prices the use of their model at $1 per use.
• The app developer believes they provide a high-quality user interface and experience, so they also set the price for each answer at $1.

If these three parties collaborate to generate a complete AI product, the total price for the user to obtain an answer each time would be $3. Of course, this is just a simple example. In reality, the pricing of the textbook needs to consider market competition factors. Market competition will affect pricing, and prices are determined by market effects.

It is important to emphasize that this pricing mechanism is not determined solely by any one platform, especially in a decentralized environment. We cannot dictate how much an asset is worth, as that would violate the principles of decentralization. Therefore, we need a framework to support and promote the operation of the market economy.

In a decentralized system, user protection and quality assurance are also important issues. In practice, our technical team uses various models for research and development. We have found that different team members have different preferences for the generation results of different models. Therefore, for subjective issues, we believe the market can self-regulate. However, for some objective issues, such as intellectual property ownership, stricter mechanisms are needed for management.

In this environment, we have introduced a token-locking mechanism aimed at increasing the credibility of participants in the network. If you are willing to lock tokens, it indicates that you are responsible for your contributions and the answers generated. For those who impersonate or infringe on others' IP, we have a strict complaint and penalty mechanism. If someone discovers that a participant has infringed on another's IP, they can file a complaint. Once the complaint is validated, the locked tokens of that participant may be confiscated or fined. Our goal is to ensure that everyone can receive their rightful economic benefits in an AI-driven world. If you plunder the economic benefits of others, we will protect the rights of the victims through fines and other means.

5. According to the roadmap, KnowledgeFi will also be launched in the second quarter of this year. As the first DApp on KIP Protocol, users have high expectations for it. Could you briefly introduce the functions that KnowledgeFi will achieve? How will this DApp empower KIP Protocol?

KnowledgeFi is planned to be elevated. We plan to develop a DApp that showcases the potential of KIP Protocol, demonstrating how to achieve knowledge monetization through the "pay-per-answer" economic framework. Currently, we are collaborating with Open Campus to explore the application of blockchain technology and artificial intelligence in the education sector. The platform will involve professors and educators, with the main idea being to provide educators with more economic revenue pathways through Web3 technology, compensating for their insufficient economic returns in the traditional education system.

For example, if a professor writes a textbook, he can upload its content to the knowledge base, and all content can be found on Open Campus University. Through the "pay-per-answer" framework, users do not need to purchase the entire textbook; they can directly ask questions to the textbook and pay the corresponding fee to obtain answers. This applies not only to textbooks but can also extend to other supplementary reading materials and articles, forming a comprehensive knowledge base.

Currently, knowledge monetization mainly falls into three categories: fixed-format sales, subscription models, and selling time. Therefore, we hope to create a new knowledge monetization model. The fourth model we propose, "pay-per-answer," is an innovative knowledge monetization model based on AI. This model does not require purchasing fixed-format knowledge products, nor does it require subscriptions or paying high lecture fees; instead, payment is made based on each question answered.

We are currently collaborating with Open Campus University to introduce this new business model into the education sector through the KnowledgeFi system. We are also discussing cooperation with certain national-level institutions, aiming to bring new business opportunities and revenue sources to the education sector through this innovative model, attracting more people to engage in education. This collaboration is significant for us, not only because it promotes the innovation of knowledge monetization models but also because it provides new possibilities and innovative pathways for educational methods.

6. Last month, the KIP Protocol Genesis NFT Mint event was very successful, minting out in a short time. The official website has also announced that Airdrop Season 2 is coming soon. Can you reveal what future community incentive plans KIP will have? What considerations does KIP Protocol have regarding partners in this aspect?

Since the beginning of the year, we have been organizing this community activity. Thanks to everyone's trust, our community has grown rapidly. A few weeks ago, we launched the Genesis NFT Mint, which received a warm response. Our Airdrop Season 2 activity is about to begin, and we have prepared a more substantial activity plan, with the scale and rewards of the second season set to increase.

We will then collaborate with Mocaverse, and specific details may be announced soon. This collaboration model is similar to our partnerships with major projects like Animoca, aiming to become their official AI partner, deeply understanding their business, and assisting them in utilizing AI technology to achieve project goals more quickly. We have many partners for the second season of activities, which will be gradually announced afterward.

7. We know that computing power, data, and models are the three essential elements of AI. Previously, you mentioned your collaboration with Aethir. How do you view the current development status of decentralized computing power networks and decentralized computing, and how do you see the future development of this track?

Currently, computing power faces a supply-demand imbalance, prompting many creative companies to propose solutions to alleviate this situation. For example, Aethir uses institutional-level computing power to centrally manage and allocate computing power from data centers, allowing ordinary users to access it. We have also recently partnered with another company that focuses on utilizing gaming GPUs (primarily those used in personal computers for gaming), which have computing capabilities similar to those required for high-performance computing.

The current market efficiency is relatively low; sometimes there is an oversupply of computing power, while at other times there is a shortage. In this situation, many companies focused on computing power have proposed various unique solutions. We believe that collaborating with these companies is crucial, especially in the current supply-demand imbalance.

In the long run, addressing the supply-demand imbalance of computing power primarily relies on improving hardware production capabilities. For instance, if Nvidia can significantly increase its output, or if a strong competitor emerges in the market, it will help alleviate the current supply-demand tension. In such cases, the market landscape will also change accordingly.

We are very willing to collaborate with companies that possess strong computing power resources. We are confident in our ability to address the computing power shortage because we know that many smart and technically skilled teams are actively tackling this challenge. By collaborating with these teams, we can jointly promote technological advancement and ensure that our projects are sufficiently supported in terms of computing power demand.

The clients of computing power are models, the clients of models are apps, and the application scenarios are realized by app developers. KIP Protocol is very focused on market development activities this year, especially in terms of apps. We realize that while data is important, our current focus is on supporting and promoting the development of AI applications. This is because AI applications are the direct clients of models, and models are the direct clients of computing power.

8. Besides the data field, will you consider entering the AI Agent field in the future? A powerful agent can receive, interpret, and execute various requests from users, while crypto provides a permissionless and trustless payment infrastructure for agents, but it also brings more complex issues. I would like to understand some of your thoughts on this?

AI Agents are a very interesting development direction. In our team's thinking and framework, we view AI Agents as a type of app. Because we can simply think of an app as a robot whose core purpose is to provide instructions to the model. Although AI Agents belong to one category, they are actually a set of independent services. We believe they should be categorized into different groups based on their service content and functions. For example, I have an AI Agent that is my personal assistant. In this concept, the AI Agent will help me automatically execute some tasks, but it does not contain reasoning capabilities itself. Therefore, in our R&D process, we will focus on specific industries, such as the education sector. We will consider what AI Agents might exist in the education sector and what business scenarios and application scenarios they might have. We will not consider it solely from the perspective of overall AI Agents, but rather from the needs and application scenarios of specific industries, as application scenarios are indispensable.

About KIP Protocol

KIP Protocol builds a Web3 underlying protocol for AI app developers, model creators, and data owners, enabling AI assets to be easily deployed and monetized while retaining complete digital property rights.

KIP will establish a new AI business ecosystem to address the problems and challenges faced in decentralized AI deployment, ensuring that everyone can enjoy the economic benefits brought by AI.

The KIP team consists of senior PhDs and technical experts dedicated to AI research since 2019, who also have a deep professional background and rich experience in the Web3 field, committed to promoting the decentralization of AI and becoming a catalyst for the decentralized AI wave.

To learn more, please join the official Chinese community: https://t.me/KIPProtocol_CN