Former Microsoft AI Engineer Guimo Sebastian on the Future of AI x Blockchain

With the advent of ChatGPT, AI is becoming integrated into our daily lives. On the other hand, blockchain is becoming more mature and its applications are increasing. What can be achieved by combining these two next-generation technologies?

We interviewed Mr. Sebastian, an AI engineer at Microsoft headquarters and a start-up entrepreneur in the blockchain domain, about the challenges and prospects in both areas, and the potential of “AI x blockchain.

Please tell us about your background up to the present.

Sebastian: I studied AI in college and joined the AI team at Microsoft after graduation. Then around 2017, blockchain started to become popular and I became interested in what was going on there, which is how I got to where I am today.

I feel that blockchain has become a very mature field compared to where it used to be and where things stand. There are now papers being written and research being done on cryptography, and it is becoming more mature.

When I was a student, I studied number theory. As an extension of that, I was also interested in the cryptography area, and that is how I gradually entered the blockchain area. As a result, I moved to a blockchain-related company in 2017, and that combination brought me to Japan.

At the time, I was involved in work related to the underlying infrastructure of blockchain. I was reading papers on cutting-edge blockchain technology and actually developing and designing new solutions.

At that time, there were many startups, not only blockchain startups, but also other kinds of startups, but they started up and soon went out of business. So there was a time when I thought that there was no point in joining a startup.

However, when I actually entered the blockchain field, my image changed. Unlike the startups at the time, they were doing solid research and development. Even if they failed, the results could be left as a paper. For this reason, I felt that it was worth participating.

Opinions on the use of ChatGPT in administrative work to improve efficiency are also questionable.

Decentralization in P2P networks is essential for AI to develop.

Sebastian, please tell us about the areas and fields where you think the use of AI is likely to develop rapidly.

Sebastian: As you know, ChatGPT is popular at the moment, but one of the major challenges is the issue of privacy. In fact, it has been banned in Italy and other countries due to privacy concerns.

In Japan, the government and politicians have explained that they can improve efficiency by using ChatGPT for some administrative tasks, but it is questionable whether this is necessarily a good thing, since using ChatGPT means passing data to OpneAI, the developer of the system. The use of ChatGPT means passing the data to OpneAI, the developer.

Since we are currently in the process of discussing specific use cases, we believe that the next step is to build a P2P network of AI that can be utilized while protecting privacy.

▶P2P

P2P is an abbreviation for “Peer-to-Peer,” a communication method in which terminals communicate directly with each other over a network rather than through a server. It has the advantage of anonymity and privacy protection.

Sebastian: What kind of use cases can you envision for a P2P network for AI?

Sebastian: For example, when the Japanese government conducts data analysis, not all data is gathered in one place, so it will be necessary to access various places. But collecting scattered data is not easy and time-consuming.

Therefore, if the data can be exchanged between AIs for analysis, not only can business efficiency be improved, but privacy protection can also be relatively assured since it is P2P. Demand for AI P2P networks should increase in such use cases.

I also believe that there will be demand in the gaming domain. In a certain paper, it was interesting to read about what would happen if each character in a village in a game world was equipped with an AI that would talk to the other characters.

However, even if a game were to be created as described in the paper, it would cost an enormous amount of money for a single company to prepare and manage the servers. Therefore, in order to realize such a game, I think it is necessary to reduce costs by utilizing a decentralized P2P network, where each company manages its own AI, instead of having one company manage the servers.

I am currently working on a project in my company called “Paima Studio,” which specializes in Web 3.0 game engines, and this kind of decentralization is one of the themes of the project. The same is true of the Shinkai Network, which focuses on AI decentralization.

In any case, we believe that decentralization with a focus on privacy will create demand.

AI x blockchain can maximize value in the area of “gaming.

What are the areas and fields where blockchain technology is likely to develop rapidly?

Sebastian: First of all, the gaming domain. There are already some Layer 2 solutions, but up until now, Layer 2 has been focused on finance, such as DeFi. However, we believe that this technology can be used more specifically for games.

▶Layer 2

A function that performs processing off-chain, separate from the main blockchain (Layer 1). This improves transaction processing speed and reduces fees.

What are some of the specific applications of this feature?

Sebastian: Technically speaking, it can be used with a technology called Sovereign Rollups. There are three types of rollups in the Layer 2 scalability solution.

The three types are ZK Rollups, Optimistic Rollups, and Sovereign Rollups.

First, ZK Rollups utilize zero-knowledge proofs, which means that the input is off-chain and not written to the blockchain. This is a privacy-preserving, computation-specific rollup, so to speak.

Next is the Optimistic Rollup, which is an on-chain input. This is a Fraud Proof method of proof, and while the ZK rollup only uses computation to prove that this proof was done correctly, the Optimistic Rollup points out problems after the transaction has taken place. This eliminates the need to verify and record every transaction, thus allowing for more efficient calculations.

Finally, Sovereign Rollup is on-chain for inputs, but off-chain for function definitions and outputs. For example, suppose that the movements of a game character are written on-chain, and everyone is replaying the movements in an off-chain local environment.

Then, the movements inputted on-chain can be output off-chain. We believe that by utilizing this mechanism and NFT, we can create games that are more tied to real-world actions, which is what we are currently focusing on. For example, it would be interesting if the NFT you own evolves as you play the game.

If you were to actually create a game, you would develop the UI and the system that is the content, but Sebastian's idea is to build the fundamental system part on-chain and the UI that the user touches off-chain.

Sebastian: Yes, that's right. Let me explain what I mean by using the card game UNO, which naturally has basic rules, but also has original rules depending on the country or region. This time, we assume that these cards and rules are output and operated off-chain.

This information can then be used to exchange opinions on the on-chain, with comments such as “I play by these rules” or “No, no, this is more interesting”. In this way, it will be easier for developers to capture the opinions of users, and users will be able to create their ideal game together.

If there is a system that allows users to exchange opinions and reflect them through applications, it will lead to an improvement in the value of games, and I think it will be important to create an environment that easily reflects user opinions in game production in the future.

What kind of value can be created by “AI x blockchain” and how do you think it can be used in practice?

Sebastian: At the moment, I think there are many blockchain game developers who are considering introducing AI because they think it will expand the possibilities. However, many people want to decentralize the blockchain if they are using it, and this is a major barrier.

In fact, if ChatGPT were introduced, it would no longer be considered decentralized. If ChatGPT were to be increasingly introduced to games, we would have to rely on developers' OpenAI, and that is not a desirable situation.

Depending on updates to OpenAI, the game itself may have many bugs. So I think the challenge is how to connect decentralized AI to blockchain games in the future, given the current demand for AI.

To tell you a little about myself, Paima Studios has developed the “Paima Engine” to address these challenges. The Paima Engine is a distributed game engine that allows anyone with programming skills familiar to developers from the Web 2.0 space to create AI-enabled Web 3.0 games, with a focus on distributed games.

To be more specific, let's say you have created a chat room with your friends. In doing so, you also possess an NFT that has not yet been touched. Then, you create a prompt in the chat room, such as “Dress for easy movement in the woods. The AI will then generate the clothes and update the NFT.

In other words, the Paima Engine makes it possible to evolve NFTs according to one's own thoughts and movements using the AI described earlier. We actually introduced the Paima Engine to various companies, and many of them were interested in it.

However, there were also many requests that if AI could be incorporated to this extent, they wanted to leave everything to AI. What they mean is that they would like to have AI installed in the game village and all the characters there, and have AI operate their actions, etc. It is not technically impossible, but it would be nice to have AI in the blockchain.

This is not technically impossible, but the cost of writing to the blockchain would be enormous, and as explained at the beginning of this article, the server management costs would also be high. As a result, the company would not be able to generate any revenue.

In this case, it is possible to reduce costs by converting the entire village to AI and off-chaining a part of it, and then creating and outputting other items in a P2P network. The part to be off-chained is the village characters.

Users are asked to purchase an untouched initial NFT and create a character by entering prompts. As this character grows, the interaction is now done by AI on a P2P network.

For example, let's say I want to shop at a store in the village. Is there anyone in the store?” and after determining its own action, it writes the information to the on-chain blockchain.

In this way, the only information written to the blockchain is the content of the action, which reduces data volume and costs. Moreover, since it utilizes decentralized AI, it can also protect the decentralized aspect of the blockchain game, which is the key point of the blockchain game.

Creating an autonomous decentralized worldview with next-generation technology

How do you want to change the world through your business?

Sebastian: Through Pima Studios, I hope to create an autonomous decentralized worldview. About two years ago, the keyword “metaverse” became a hot topic. There have been many fundraising cases about realizing the next generation world, but most of them eventually collapsed.

The reason is that it is difficult to create a decentralized metaverse with today's technology. Given the current progress and other factors, I think we will need even more decentralized metaverse in the future.

Pima Studio also hopes to create such a decentralized world in the future. We developed the Paima Engine about a year ago and have been releasing games, and we are gradually moving into that phase.

With this future in mind, we are currently focusing on creating an “on-chain RPG” - a game where you catch and play NFT-ized monsters. We plan to incorporate elements of creating villages and quests in this game, allowing each player to create their own “world”.

We also want to use AI to create Play to Earn, where the ecosystem does not collapse. There are many challenges, but we want to design a game that allows users to enjoy the game world and earn income while contributing to the game.

Profile

Sebastien Guillemot

Sebastien Guillemot is co-founder and CTO of dcSpark and co-founder of PaimaStudios. dcSpark is one of the world's leading crypto ecosystem architects, developing many new projects for the global blockchain ecosystem, including Flint wallet and Milkomeda. PaimaStudios, founded in April 2022, is a leading company at the forefront of the industry's efforts to create a trustless Web 3.0, focusing on building web primitives and pioneering the PaimaEngine, which is expected to become the new standard. PaimaEngine is expected to become a new standard.

Related Articles

An Industry-Specific Efficiency AI App That Could Change Japan

AI PROMPT Mastering the Magic of Modernity