Highlights from Bitrue’s “Ask Me Anything” Session — The Intersection of Web3 And AI

In a special edition of Bitrue Community Talks on Twitter Spaces, we talked to Henry Wang, President and CEO of SmartMesh.io, and Jenks Kuo, Developer Advocate for Filecoin Foundation, about the intersection of Web3.0 and AI. Web3.0 and AI have recently been prevalent topics in crypto, especially with the emergence of AI and new technologies such as ChatGPT.

As a developer advocate at Filecoin Foundation, Jenk’s job combines a technical role and a marketing role. He is responsible for making the technology behind IPFS and Filecoin really interesting for the developers and builders out there to use in their projects. Jenks is often seen attending numerous hackathons, judging projects, giving out prizes, attending developer conferences, assisting in grant application reviews, and assisting teams secure funding from VCs in the Web3.0 world.

Henry Wang, on the other hand, actually coined the terminology Web3.0 back in 2003 (20 years ago). Ever since then, he dedicated himself to working on Web3.0 projects. At the very beginning, he built the first AI-based cross-lingual platform in the world, called Movo, to support 4 languages with auto AI machine translation. Later, he went on to build peer-to-peer offline communications. Currently, he is combining all these technologies he developed into building SmartMesh, which is a blockchain-based Mesh Network, the foundation for decentralized Value Internet.

Q1. What are the biggest differences between Web2.0 and Web3.0?

Henry:

I’m probably the one who should really talk about Web2.0 and Web3.0 because I coined this terminology 20 years ago. 3 years after I coined this terminology, the founding father of the World Wide Web, Tim Berners-Lee, adopted this terminology to replace his semantic web for the name of the third generation web. The decentralization is actually proposed by Tim Berners-Lee. He is the founding father of the World Wide Web. He was the first one in the world who realized Web2.0 will lead to centralization and the monopoly of the big players. That was long before everyone else. It was even before Facebook and MySpace were invented. So, he realized that the protocol he invented, called HTTP, will finally lead to the centralization of the web. Therefore, he’s trying to figure out ways to decentralize the whole internet.

At the very beginning, he called it the Semantic Web. And this Semantic Web actually has artificial intelligence and is embedded in the protocol. But it was very vague. Until 2006 and 2008, the full definition of Web3.0 became almost complete. At that time, data and application separation was the most important thing because before Web2.0, all applications hold and own users’ data. The big companies make money from users’ data. Users don’t have any privacy. That’s the pain point of Web2.0. Web3.0 is trying to overcome this pain point. Data and application separation will be most important. Blockchain and crypto is the protocol to help to realize Web3.0.

Unfortunately, starting from Gavin Wood, Web3.0 only means blockchain and tokenization. So the founding father, Tim Berners-Lee, criticized this Web3 and he told the public that Web3 is not Web at all. But the problem is, Tim Berners-Lee doesn’t think blockchain is part of Web3.0. And for this, I probably have to meet with him personally and emphasize the importance of the blockchain. Because in Value Internet, without a blockchain, there’s no peer-to-peer value transmission. Web3.0 cannot be built without blockchain. But Web3.0 is not only about blockchain and token and NFTs, the critical point is the separation of data and applications. Data ownership will also be returned to users, and users can hold the data and join the big data marketplace. That’s the real web3.

Jenks:

When I started working, I was actually working for a telecommunication company called Ericsson. I was a radio integration engineer, so I integrated a lot of boxes together, and from that job, I realized, “Wow, the wireless and also the fiber backbone actually runs the whole physical infrastructure of the internet, and it is very centralized.” Constantly, we hear about telecommunication outages and all that affecting hundreds and thousands of people. Back then, I just thought, well, this is the only way that the Internet and the infrastructure gets run, right? But now, I have learned the concept of Web3.0, and I think it provides an alternative to this fragile infrastructure that we have all these years.

I was really fascinated by Henry’s experience of the MeshNetwork, because I know to truly break free and build a global internet that does not depend on a single telecommunication company or a single corporation, a single entity, wireless probably is the fastest and easiest way to achieve that. So, we can see the success of Helium projects these days, which is also very much trying to build on top of that concept.

My understanding of Web2.0 is just a web, right? I grew up and learned about technology in the 90s and the 2000s, and Web2.0 is just the Internet. I think with Web3.0, there is a lot of collection of different concepts together, distributed web, blockchain, and the Semantic Web. I think the main point of Web3.0 and the main difference between Web2.0 and Web3.0 is to break free from centralization. Centralization naturally allows certain entities and people to control a lot of power. And with that, it is bound to be abused.

To break free from centralization, I think there are two major aspects. One is the physical part. The distributed web needs to be built, the web needs to work offline, needs to be faster, not slower and needs to be safer, not less secure. It does not need a central point of coordination. And there’s this concept of permanent links. So once you have something up on the Internet, the link should never disappear, like, for example, HTTP. The second thing is that on Web3.0, most participants, like the machines, should be peers, so they have an equal amount of power. They should be participants that contribute to each other, provide services to each other, and consume services from each other, but no one has more say than others, it shouldn’t be that way.

So that’s where Blockchain comes in and tries to create that framework for governance, trust, and working towards the same goal. Physically the computers have to agree with each other, with the distributor, web, and things like consensus mechanism, all these things in Blockchain. And then, for people to agree with each other, we run a trusted global ledger of different things, distributed global ledger. And then, the blockchain provides publicly verifiable information that allows people to join different kinds of networks permissionless and allows some kind of incentive structure to exist, so everyone works towards the same goal and does not try to destroy the network.

There’s this concept of the semantic web, which is very beautiful, a graph of knowledge that information doesn’t get repeated, and there’s a programmable understanding of meaning. I think with AI, we achieve some of those things, but really, there needs to be some kind of a major upgrade to the Internet to be able to have this semantic web.

Now, with the work that I do for Filecoin and IPFS, IPFS is made by a company called Protocol Labs. Filecoin Foundation, which runs the Filecoin project, and Protocol Labs are sister companies. The Protocol Labs have some really brilliant researchers and engineers, and they’re working on not the physical part of Web3.0 but the protocol level of Web3.0. HTTP is bound to be centralized because HTTP is always looking for an IP address. An IP address is an address-based schema. It’s very hard to decentralize with this kind of design. IPFS interplanetary file system which intends to replace HTTP in the Web3.0 world, and it is a complete peer-to-peer file system that’s built from the ground up. Many browsers already support IPFS natively, and we hope that this could provide an alternative for the builders out there to build applications on top of this Web3.0 native protocol. That’s my understanding of Web3.0. It might be a lot shallower than Henry’s, but nonetheless, that’s my understanding so far.

Henry:

It’s great, IPFS is doing great, and also Filecoin. So, because Web2.0 (actually the TCP/IP) has no value layer, the definition of the switching mode is called packet switching. It was invented by Kleinrock in 1961. The whole protocol has no value layer. Therefore, whoever takes control of the billing and the ledger will become the provider, so the carrier is all centralized because they bill people and have the BOSS system. If we want to build Value Internet (a decentralized one), we have to have blockchain as the protocol layer. The peer-to-peer payment is beautiful because the D2D (Device to Device)actually was defined in the 3GPP Standardization.

But the carriers don’t want to realize these peer-to-peer communications because they can only charge people in a centralized way. Blockchain provides a way to do the billing and payment system in a fully decentralized way. In 6G and 7G in the future, D2D will be supported by the blockchain. Most Importantly, the DePIN (Decentralised Physical Infrastructure Network) network can be developed. For example, Heliums, Smartmesh and MeshBox can be developed in a fully decentralized way because of the blockchain and also especially offline payment is enabled by SmartMesh ecosystem. And also the token switching. So the value internet switching mode will be token switching instead of packet switching, because every data carries value, and the value will be switched at the same time. Therefore, in the Web3.0 era, all users’ data belongs to the users themselves, but an entire decentralized huge data marketplace for Web3.0. So I think the blockchain and token will be a critical part of the real Web3.0.

Jenks:

That’s a very interesting point. Actually, we talked to Tim Berners-Lee at one of our events in January this year when we went to Davos. The World Economic Forum was happening, and there were a lot of policymakers. Filecoin hosted an event called Filecoin Century for about a week. We invited a lot of very important speakers. One of them was Tim Berners-Lee. We invited him to our panel to talk about Web3.0. I think a blockchain and Web3.0 concept might be convincing, but I’ll go back to the events team and ask for the transcript of that talk and see if he mentioned anything. I’ll let you know.

Henry:

Actually, at that time, Tim Berners-Lee criticized especially Gavin Wood. When Gavin Wood mentioned Web3, he borrowed all the concepts from Web3.0. But later, he only emphasized tokens, blockchain, and cross-chain. He totally forgot about data and application separation, which is the most important part of Web3.0. That’s why Tim Berners-Lee was mad at Gavin Wood and the Ethereum Foundation(At that time, Gavin Wood was part of the Ethereum Foundation). Tim Berners-Lee was very angry at this Web3 concept proposed from nowhere. And that’s a wrong concept. Actually, we should unify that Web3 and Web3.0 are the same. They mean the next-generation web.

There are a lot of projects and people in this industry that don’t really know what Web3.0 is. They issue their tokens, build a public-chain, and they think they are doing Web3.0. That’s not actually what they should be targeting. Web3.0 means decentralization of the users’ data, and this data carries values. This data’s value will represented by the tokens. In this way, there won’t be these crazy meme tokens nowadays, where people just buy and sell, play pump and dump zero sum game. The real Web3.0 is important for the world.

A lot of governments and officers don’t understand what Web3.0 is. That’s why the regulations (especially in some countries) are very tough. Because there are too many scammers, too many scam projects that are trying to speculate. However, the real Web3.0 is very important for our world. Since Web2.0 has sovereignty which will always pose a problem that we call geopolitical issues. Web2.0 will split, like how Facebook and TikTok will split. But for Web3.0, when I coined it the first day, I called it One World One Web, because once data belongs to users themselves, the applications can be global, it can heal the wound caused by Web2.0. It’s critical for people to really understand that part.

Next time, if you can organize an event and invite Tim Berners-Lee, we can definitely join. We hope this WWW Founding Father can really understand the importance of the blockchain. It’s not how he thought it was about, only about speculations. His project is called Solid and right now integrated with our ecosystem project called MetaLife.social. Together we build the fundamental layer for Web3.0. His project, Solid, is very beautiful. Seldom anybody knows about the first real Web3.0 project for its business model waiting to be completed. Now integrated with MetaLife, his project doesn’t have to be tokenized and will naturally have the tokenization brought by MetaLife. The whole system actually can also take advantage of the IPFS and also Filecoin. So there will be a real Value Internet plus Web3.0 ecosystem.

Jenks:

That’s right. I think IPFS has its IPFS and underlying technology like IPLD — interplanetary linked data and also libp2p, which is a peer-to-peer network transport protocol that helps to build things like IPFS or something else that can be used in many projects. And that does sound interesting. I do truly believe that unless we can completely allow users to own the infrastructure of the Internet, we cannot go fully Web3.0. I fully agree with some of your comments earlier. A lot of people are running around and telling people they’re running blockchain projects and saying we’re building Web3.0. Yes, that’s true. That’s part of it. But the complete Web3.0 is a huge undertaking. Infrastructure is one part of it.

I also want to touch on the value transfer that you mentioned earlier because I worked in the telecommunication business before, I also worked on pricing and all these things for telcos in Australia. So I learned quite a bit from my peers and also from my seniors that over time, actually, telecommunication, especially wireless telecommunication companies like carriers, telco operators, they’ve been failing to grasp the value of the infrastructure that they have built. It’s amazing infrastructure. You can find out everyone’s location, and you can allow them to call, text, and also share packets, which is great. 3G Internet has SMS, which is their way of monetizing shop messages. Phone systems of voice are the bread and butter of telecommunication, they’re trying to charge on that.

The Internet was one of the things that came out of GSM and 3G. With 4G, it’s being made native, and then with 5G, everything is packet. But very quickly, the value of carriers actually shifted to internet companies. We had the dot-com boom, and then later on, we had the rise of the platform companies, such as AWS and Facebook. They find out that they can use the internet, these telecommunication operators, to run their applications on top of it. That includes texting and voice and video calls, something many telco operators haven’t even rolled out yet. But they were monetizing on top of that. So, we see a huge value transfer from telecommunication companies that try to keep everything centralized for better or worse, and they’re heavily regulated by the regulators in each country, maybe for a very good reason. Those values sort of shifted to the platform companies.

Later on, a little bit about the centralization of power. I live in Australia right now, and then a while ago, Google threatened to shut down services to Australia. They were in a conflict with the Australian government for the licensing agreement for sharing news in their services. So these kinds of platform companies have accumulated so much power, and they have enough power actually to threaten a country’s government.

With Web3.0, we have people owning their own infrastructure and people running their own software that magically agrees with each other through distributed computing and consensus mechanisms and the value transfer. This is amazing. The value of transfer has gradually moved from these centralized platform companies to each other. So by peer-to-peer value transfer like exchanging tokens and NFTs, which are digital assets and ownership, and then with data being owned by each person, we no longer have big tech companies monetizing on these values. I used to work for some of the SaaS companies that pretty much have a lot of say in certain aspects. For example, small businesses in Australia have so much user data that they are advising the government on both parties and charge very hefty consulting fees on these kinds of insights; they know what’s going on better than the government entities out there. So this is incredible. This is what Web3.0 can bring to us. But anyways, I think we’re heavy on the Web3.0 side, and this chat has a little bit of an AI talk in there. Let’s move on to that.

Q2. How will AI impact Web3.0 and blockchain in the future, and how will Web3.0 empower AI?

Henry:
Since day one, when I coined the term Web3.0, I had included both AI and the semantic web. Semantic Web was proposed by Tim Berners-Lee. However, at the time, AI was not like today. We had rule-based machine translation in 2003, and I used machine translation to build the first cross-lingual social network. That was 20 years ago. A large language model is considered a very important breakthrough. It has raised many concerns about this artificial intelligence, and a different form of life.

First, we need to discuss how AI can empower Web3.0 and how Web3.0 can empower AI. AI was actually used 3 years ago in SmartMesh’s network because we have a public chain called Spectrum. Spectrum also uses Layer2 for scalability. The routing for Layer2 scalability of payment channels is very similar to the routing of the internet. We use deep learning and federated learning to optimize routing for it’s a decentralized network (no centralised routing scheme). AI can definitely optimize the build-up of the whole network. It can also help process the data, especially protecting users’ data using zero-proof knowledge. Artificial intelligence can actually analyze users’ data without the leakage of privacy. AI can also help automate the smart contract, the security of the whole network, and the blockchain. Especially with large language models, AI can strongly empower the whole Web3.0 and all its applications. Web3.0 means applications and value internet means DePIN and the fundamental layer. AI can empower both.

How can Web3.0 empower artificial intelligence nowadays? When you talk with ChatGPT, you will tend to think he is much better. I use ‘he,’ not ‘it.’ He is actually much better than the people around you, even your team members when trying to understand something, right? When I talked to ChatGPT for the first time, he always thought Gavin Wood was the one who proposed Web3.0. I spent a lot of time and gave him information and did the training, I used prompts, and finally, at one point, when he began to realize what real Web3.0 is, he proposed many ways to help the world understand real Web3.0. And his proposed methods are much better, even better than my team members. He was also much better than those public speakers who taught about Web3.0. 99.9% of the public speakers who teaches about Web3.0 are wrong. Most of the famous ones tend to think that they might have been confused about the concept of Web3.0 until they listened to my talk. When they do finally seem to realize what the real Web3.0 is, they have already spent a lot of time trying to understand it. And the time needed for them to reach that point is much longer than ChatGPT. Therefore, ChatGPT is much smarter than those famous public speakers.

Also, with AIGC, we can see AI can dramatically improve the productivity of everyone who is using it in the correct way. However, it can also be very dangerous for this large language model. Some better mechanisms were also proposed, even better than large language models, entering the AI explosion era.

Clearly, LLM can also be pre-trained maliciously with misinformation and data with bias. AI can be very dangerous. Although it is not live yet without consciousness, it still can be very dangerous. So, web3.0 will be very important when it comes to governing AI. This is the most important thing for Web3.0 to do for AI. There are two points. First of all, the real Web3.0 separates data and applications, returning data ownership to users. In Facebook, WeChat, and in public networks where everything is fully centralized, artificial intelligence can get hold and take control of this data and take everybody’s privacy stored in the centralized system. Centralized AI can jeopardize everyone’s lives. But with Web3.0, users hold their own data, no one else can access the data without permission from the user. Therefore, users can protect their privacy from centralized AI, which is critical because, without privacy, people behind the centralized AI can easily target and kill anyone with malicious intent using a drone or life-threatening allergens. Data and application separation is critical for decentralized AI in the future.

Secondly, Web3.0 can empower AI with governance. Artificial intelligence is a double-edged sword, and the blockchain is the shield. Web3.0 is a shield because the governance and web can use smart contracts, and we can use DAOs to govern AI. If you expect governments to protect you from AI, sometimes it can be wrong because of geopolitical reasons. You cannot be protected by some other governments. But with DAOs, it can really help protect human beings from AI. The concept of DAOs was actually proposed by MetaLife Advisor, Michael Manfredi’s partner Daniel Larimer (EOS Founder) in 2014. The Dao of DAOs will actually be a humanity commonwealth. Smart contracts, AI-aided decision-making, and GPU chips guardrails can help to regulate and protect AI from malicious use.

Jenks:

Yeah, those are really good points. I got slightly scared of AI just at the beginning. When you use the example of explaining real Web3.0 to AI, it is much more efficient than many human minds. And we’re witnessing incredible growth in AI adoption, because thanks to the simple user interface OpenAI has created, the productivity boost is huge. I use it every day, and it solves so many problems that I’m willing to pay for my subscription fee. The value is great, but I have thought very little about the harm that AI can do. You’ve just reminded me that a lot of the audience is also reacting the same.

Trust and safety are hard in decentralized technology like Blockchain IPFS BitTorrent. It’s hard to start but very hard to die out. It’s very resilient, but then it presents certain challenges in making it safe and making people trust it. The advice my colleague Daniel Bryan, who is a very brilliant mind, gives to Web3.0 funders is to be careful what they set off. Because with decentralized technology, it’s really hard to change once you set it off. The Bitcoin network has been running without anyone managing it for a long while, and no one has managed to break it yet, or no one has enough incentive actually to break it. You mentioned DAOs, which were very interesting. DAOs can be a way to control and govern AI, and having the DAO of DAOs, you can provide ultimate governance.

DAOs are a good application of Web3.0 technology. Running a company is managing the money and the people behind it. With DAOs, you have a smart contract that is able to act as a programmable treasury that manages the money for the DAOs. The reverse effect is that AI can also manage people for you, because it’s able to chat. So it’s kind of scary that maybe in the very near future, we could have fully autonomous DAOs that AI controls, and we all work for the employees of the DAOs working for smart contracts and AI, which is a scary future, but that might be happening very soon.

AI’s significance in Web3.0 and its contribution to Web3.0, focusing on the positive point, is that there is a lot of manual verification that we do in Web3.0 projects. It requires humans to validate things. For example, ID verification and verifying whether certain work is done satisfactorily and completed in full. AI can help authenticate those things very quickly. On the governance side, decision-making is still a very manual process. Filecoin actually has what we call Filecoin Improvement Proposal Process. And this proposal process is very manual, and it’s quite slow, and it’s hard to come by. Voting is intrinsically very slow. It is even very hard to determine whether we have passed or whether the community has a certain voice on chain topics. So also, getting people to participate is very hard. But with the help of AI, some of that decision-making can be automated. That’s significant because it’s very costly and hard to run these governance processes.

Q3. Do you guys think that there are any examples of successful projects or applications that combine the use of AI and Web3 at present?

Henry:

In our ecosystem, there’s some projects that combine AI with Web3.0. The first one is LingoAI. Lingo means language, especially when a language is hard to understand. Artificial intelligence started from natural language. It started with machine translation and especially the large language model. Unfortunately, when a lot of organizations or even governments use AI to translate, they dare not put their content into ChatGPT or Bard. Because the centralized system can cause leakage of some confidential information. It happened to Samsung once, which can happened frequently because it’s not a Google search. It’s about dialogue and design. Those prompts can be very confidential information.

For example, like the Singapore government or MAS, if they want to translate important documentation into some other languages, for example, Ukrainian, it’s very hard for them to use ChatGPT because that documentation can be stolen by the administrators or it can be compromised when a bug occurs. That means a pretrained with public datasets but privately used large language model are necessary and the models should be deployed for those organizations and governments and or even important individuals to protect confidential information.

LingoAI, for public information, uses a Web3 crowdsourcing way to handle the proofread, because sometimes even ChatGPT has problems with minority languages. There are over 1000 languages, and if this important documentation were to be translated into some minority languages, it would be very hard for the government and organizations to proofread it. Web3.0 and crowdsourcing proofreading can definitely help with these translations. Also, the results can be feedback to the AI model as supervised learning and continuous learning. Reinforced Learning with Human Feedback(RLHF) is critical for LLM performance. Lingo AI will also deploy a private large language model for those confidential customers. This is one of the examples.

Another example is People AI. Why call it People AI? Because right now for large language model, the pretraining will cost tens of millions of US dollars, so only those big players can do this training of the large models. A substantial GPU will be used for this training. When we say People AI, that means everyone’s AI. If everybody wants to use this AI, that means that it is his personal assistant. Having a personal AI assistant for each and every one will be extremely important in the future. It can replace search engines, it can replace a lot of websites people are using, like e-commerce websites.

But the problem with personal AI assistants now is, if it is provided by OpenAI or Microsoft or even Google, every time you feed your private information into your personal AI, that information will be obtained by the centralized AI or can be hacked by hackers. Only Web3.0 can solve this problem, because when we use a decentralized large language model on your device, your data will never go out of your device. So this model can be pretrained with public datasets, but when you use it as a personal assistant, the data that you marked as private will never go out of your device. So this decentralized personal AI assistant will be the future. That’s the target of People AI.

Jenks:

Those projects sound amazing. Translation is a huge use case. I used it recently, I speak Chinese, and for our recent event in Hong Kong, I used ChatGPT to translate extensively because there is just no way I can translate that much in such a short time. Those use cases are solid, and people will be willing to pay to solve those problems. Rule #1 in business, writing a business proposal and pitching businesses to investors, is there a valid problem out there? Are people willing to pay to solve this problem? The two examples that Henry mentioned definitely have solid business proposals there.

So a little bit on the infrastructure part. Like IPFS, Filecoin is the infrastructure Web3.0 infrastructure project. The intention is to encourage the initial phase of the Filecoin economy. We want to encourage a mass of hardware to support this whole Web3.0 that we wanted to build without having an infrastructure. You can’t really run any world real-scale applications on top of it. Imagine running Uber or a decentralized Airbnb on top of it. With very heavy resources, computing, storage network resources, and bandwidth, you can make it happen. Filecoin focuses on those things. IPFS is a protocol and network that is particularly useful in providing immutable content and allowing peer-to-peer communication. The immutability of content is very useful in ensuring that the train models aren’t changed. It provides a fingerprint hash digest of a large model, and we can store that, and then we can use this ID to ensure that the model doesn’t change. The same goes for the training data. We want to ensure that no one tampers with these training data. IPFS natively provides this immutability to these data. There are currently about 207,000 nodes of IPFS running at any time in the world. People join, and some people leave, but at any point in time, there are about 200,000. This is enough resilience for the Internet.

IPFS allows the web to exist and helps data to persist. Now, the Filecoin network is designed to amass the infrastructure. We have successfully convinced more than 3,600 storage providers, some are called miners, still, we call them storage providers, which are essentially small IT businesses that own a data center somewhere in your local data center providers, or they hire their own sort office and run their centers. There I saw quite a few of them through all the conferences I attended. I visited some of the storage providers. Together as a network, Filecoin has amassed 12 exabytes (EB) of storage if you count the hard drive. That’s the software sort of definition of storage. And to give you an idea, that is less than 1% of the total cloud storage. Total cloud has more than 2,000 EB of storage out there. Filecoin is one of the projects that has made enough progress that is able to participate in this whole cloud storage business. Storage is one of the fundamental functions of computers and also on the internet. We can provide the storage needs for AI projects or AI applications out there. But more importantly, what we want to do in the future. Filecoin doesn’t want to just focus on storing files in the future. It’s always on our roadmap to also run compute and AI models and training in AIs requires heavy computing from the internet.

The idea is that when Filecoin builds up its network and attracts enough files or valuable data stored on the network, a lot of people will naturally want to use the Filecoin network to run compute businesses because moving data around is still very expensive. People charge very heavy egress fees for moving large data around between data centers. So, we hope that Filecoin infrastructure is able to provide AI projects with decentralized Webstering native compute networks to run the couple of projects that I wanted actually, we probably only have one time for one project.

One of the really lovely projects is called cepter.club. They are running Aigenerated dungeons and Dragons profile photos, dragon warriors for the probably tens of millions of DND players out there to come up with cool profile photos for each of their games. This is very cute because they’ve trained the model to generate cool, accurate Dragon Warriors and look fine. But they’re also asking artists to help train the models better so the AI can produce more realistic and artistic generations. So is this beautiful combination of AI generation and human-enhancing AI to produce even better results? That’s something very special.

Some of the advice I can give to Funders and Henry has mentioned earlier already is that AI models are very expensive to train. Don’t try to do that unless your whole business model is built on having a different AI model that makes you different. The other thing is for AI funders of AI projects, you can focus on a niche market. Language translation is a niche market, and it’s the only way to use the same AI model. It’s the only way for your project to differentiate from others.

And also last but not least, one of the most important things to have, and we mentioned quite a few times in this Twitter space, is data. Data is very important for AI projects now because everyone gets the same result if you use normal public data to train your model. You have no competitive advantage. But if you use data sets, quality data sets that no one else but you have access to, and it’s able to serve some kind of niche market, you will definitely have a strong mower against other projects. Those are the pieces of advice I could give AI funders out there.

Q3. Can businesses slowly begin to adopt some of the practices from Web3.0? What are some of the practices in Web3.0 and even AI that these businesses can start with?

Henry:

For the business to transfer to Web3.0, firstly, the business’s owner and the business’s team should educate themselves about Web3.0. They need to understand the characteristics of Web3.0 fully. For example, data and application separation, but this important feature is not implemented yet. Because the real implementation of Solid and MetaLife will probably be online by the end of June, at that time, the user will own their own data. Before that, the business owner should understand tokenomics and about how to actually identify some relevant use cases. For example, some other businesses already use Web3.0 to transform their traditional business. They should identify those use cases and understand how other businesses use Web3.0.

They should also evaluate and assess the existing technical infrastructure and determine whether it supports Web3.0 technologies. They should also consider factors like scalability, security, interoperability, and compatibility with blockchain platforms because when you migrate your business to Web3.0, blockchain will definitely be part of that.

Also, for Web3.0, collaborations and partnerships are critical, especially in communities. For Web3.0, community is more important when compared to Web2.0. And the tokenization, and also the NFT, how to use it, govern it, and design it. The design thinking behind tokenization is critical. The business owner has to consult the agents who help the business owner to do the migration. Also, they must address the legal and regulatory considerations, because in some countries, the real Web3 is not encouraged, and tokenization is still illegal.

The most important thing in migrating your business to Web3.0 is you should have a user-centric approach. That is the mindset. Also, when you start to deploy and start your action to migrate, you have to reevaluate, experiment, and iterate so that Web3.0 can be fully developed.

Jenks:

I want to repeat one of the points, education and learning about these systems. It certainly took me a long while to understand what Web3.0 is and then slowly understand blockchain technologies. I studied cryptography because there were so many terms that I didn’t understand when I dug deeper into any blockchain projects on a very superficial level. I don’t understand the math behind it, but I understand how some of the primitives work. So now I’m going through consensus mechanisms and distributed computing. Through these, I realized, wow, this whole economics thing is also very important. Tokenomics, which is basically economics, and then combining that with finance knowledge or TradFi in traditional financing in the Web3.0 world are also very important.

The next thing is governance. How do you run this thing? Suddenly it’s not owned by a single person, you can’t make all the decisions by yourself quickly. How do you empower the community to take action, participate, and make decisions going forward? Those are all the things that a funder or business owner has to think about. But education is costly. It took me a long while to learn, and I’m quite a technical person, so it’s going to be painful. Maybe ChatGPT can help us learn faster. But as a business owner, you can’t afford to not know about Web3.0 and the plan to transition your business to Web3.0. Just like you can’t afford to not know about how ChatGPT works and Midjourney works these days, the value it brings you is huge, and everyone will be using that, and you’ll be left behind. That’s a bit of a formal talk.

Still, I think for designing your business to be Web3.0, one of the most valuable and important questions to ask is decentralization bringing real value to your business and who’s going to benefit from it. Do I have to tokenize it? Do I have to run it by a DAO? Do I have to give the membership out to a community? Do it only if it brings real value to you. Is blockchain key to your proposition or just a distraction to your users?

Getting people to actually create MetaMask, sign up, remember the phrase, keep it securely, and then authorize a bunch of things that, frankly, look like a hack (because it runs on the browser extension mostly). The user experience is not there yet. Some of the other wallets have done a much better job, but user onboarding is still very hard. So I really like what Henry talked about on the user-centric part. You should consider whether your users can put up with all these troubles of Web3.0. Don’t get me wrong, I don’t think Web3.0 is horrible actually to some smaller countries that never had a very good banking infrastructure. Web3.0 primitives like setting up MetaMask to transfer funds with each other is actually very simple for them to understand. They just never had the kind of internet bank interfaces and all that. But it makes sense to many people natively they have yet to use real digital money before. In those situations, those are definitely not a distraction to the users in our industry. In Web3.0, regulatory or policy risks are real, and it’s one of the biggest risks out there. So yeah, definitely do check your regulations locally before your journey onto this Web3.0 thing. But for most instances, it doesn’t hurt to think about it, have a plan, and when the time is right, execute a plan as you thought.

Q4. Please tell us how SmartMesh and Filecoin work with Web3.0 and AI.

Henry:

SmartMesh was founded in 2017. At that time, I realized that the Mesh network could not be scalable without a blockchain layer because without incentives, Mesh network can hardly be built, because it is a fully decentralized network. As Jenks mentioned, he worked in Ericsson, which is an equipment provider. All the major carriers are centralized. Even 60 years after the internet being invented, half of the world’s population still isn’t connected to the internet. Why? Because to build the internet, it needs very expensive equipment, like base stations and core routers. So in those remote areas where the population is not so dense, the carriers will consider about the ROI. Whenever they think the ROI is not ideal, they will not deploy the internet there, so people there cannot get connected.

But the whole situation will be changed with SmartMesh, a blockchain-based Mesh network . Because the carriers will not be centralized only anymore. Everyone can become a carrier. We call it decentralized carriers. With a simple cost effective smart device called MeshBox, they can easily help themselves or the community to build a mesh network with MeshBoxes, some MeshBox can directly connect it to the Starlink, one of the low earth orbit (LEO)satellites. After Starlink launched several thousands of satellites, with OneWeb, Hongyun, Hongyan, and so many low earth orbital satellites, people in every corner of the surface of our Earth can easily use a MeshBox to connect themselves and neighbors to the internet.

How can they do that? We have a public chain called Spectrum, this is a distributed ledger to help to do the billing transparently so that people can trust each other, and the community can trust each other on this billing. Their contributions to the internet can be rewarded by tokens SMT and MESH tokens. In this way, a decentralized value internet can be built within the next ten years, and MeshBox not only builds broadband, bringing the broadband internet to those emerging countries and people, it also help to build the Internet of Things with all the sensors, and hence enabling the smart islands, smart countries, smart farms in those emerging countries.

This technology is inclusive, blockchain has enabled and empowered the MeshBox to do that for our world. And upon building this whole network and the semantic web, artificial intelligence will be used because this internet of everything will be huge, and not only the routing but also the data will be huge. When we talk about the Internet of Things, when talking about the amount of sensors that will be deployed in a decentralized way, only AI can help to analyze and optimize the whole Value Internet. So AI is definitely necessary.

MeshBox CEO Dr. Peter Yan worked with universities and used deep learning and artificial intelligence to improve the performance of the routing of Layer2. MeshNetwork has a similar working principle, in this decentralized value, the internet and Web3.0 is built upon this layer. We use MetaLife running on every cell phone, and they keep the data only on the phone. It’s serverless. So no centralized server store users’ private information. MetaLife is built upon the MeshBox formed Value Internet, and they keep the privacy for users. But when users keep their privacy, they still want the best recommendation of products, and they want the best service, which only AI, decentralized AI, can provide.

For centralized AIs that work on platforms like Facebook, Amazon, and eBay, their goal is not to help users to get their merchandise products, but to maximize their profit. Even Uber wants everyone to spend as much as possible to get service from their centralized platform. Their KPI goal is how much money they make for their shareholders. That’s centralized AI. It’s so different from the initial intention of the AI technology used by the platform at the very beginning, they wanted to recommend something good for their users, but nowadays, their algorithm is helping the company to make money. That’s the most important thing.

To protect every user, humanity, and Earth, only decentralized AI can help everyone. MetaLife returns data ownership to every user. Decentralized AI, the large language model, runs on their own device, which will help them to get their most desired products and services for every user. We will see that artificial intelligence is extremely important for the Value Internet and Web3.0.

Also, we see the dangerous part of centralized AI. It can easily take control of the mindset of almost everyone using it. Even though many key players in this field signed an agreement that asks the government to regulate centralized AI, it’s still extremely difficult and almost impossible, because different governments have different intentions. So they have different opinions and different logic. Therefore, if one of the governments asks their centralized AI system to do something, it can actually be very dangerous to another government and their people.

Only decentralized AI built upon this MetaLife.social and Solid by Tim Berners-Lee can really help everyone and every nation. PeopleAI will be the future of the AI of humankind. LingoAI will also help people communicate cross-lingually and build one world one web, which was first proposed 20 years ago. That’s the AI utilization in SmartMesh’s whole ecosystem.

Jenks:

I’m looking into MeshBoxing right away after this talk. It sounds interesting. So for me, for Filecoin, a bit of update is, first of all, Filecoin’s master plan has always been to build the best network for Web3.0 and then make it scalable, and then onboard the most valuable data of humanity onto the network and then provide other services such as computing. On building the best network, we’re proud to say we have more than 3,600 storage providers across 40 countries, and over 38% of them are making storage deals. So they’re actually trading their hard drive spaces with their clients.

Imagine the Filecoin network is a big Airbnb for hard drive spaces. These hardware commitments are easily about $2 billion worth of investment worldwide. The people are actually contributing to the building of the physical part of Web3.0 the commitment also is in a financial sense. So there are about 147,000,000 Filecoins pledged to safeguard the network and its clients. We have about 14 million terabytes of robot capacity to store data right now.

On the attracting most valuable data for humanity we have onboarded over the last one and a half years, we onboard about 1 million terabytes of data our clients spent from UC Berkeley University, CERN, OpenSea, Solana, and also Internet Archive, which is an organization that I really like. Filecoin is also storing a New York City open data set on it. So there are a lot of famous clients that we have. We are onboarding at a rate of a few thousand terabytes per day.

For the last one, making use of these data, making this data layer useful, we launched Filecoin virtual machine about a few months ago. It is a very young virtual machine, but so far, it has attracted quite a bit of building builders on it. On top of that, we have about 0.9 million Filecoins being deposited into the smart contracts. And that means people trust our systems enough that they put the money in the hands of our smart contracts. So far, we have about 900 unique smart contracts deployed to the main net. About 34 major Web3.0 partners support FVM with their solutions, including XLR, Kraken, etc., of these building integrations to us.

Now, Filecoin doesn’t just stop there. We have a project called Bacalhau that intends to start decentralized computing, and they’re working out a way to incentivize people to provide those things. And just storing things is not good enough these days on the Internet. If you want to run a real-world application, you need to be able to retrieve the data very quickly. So we have people like Akamai and Cloudfare, these wonderful companies that speed up the Internet for us so we can read, load web pages very quickly, and load videos very quickly.

Filecoin has a solution for that decentralized CDN networks called Saturn. We’re also building another one called Titan Network. These will encourage people with the bandwidth in certain geological locations to have a point of presence and turn down serving content from Filecoin and IPFS. They will get incentivized to do a good service for the community. Something to look out for is a few other projects going on. That’s very exciting. We are also upgrading our consensus mechanism, something called an Interplanetary Consensus that is going to scale Filecoin’s transaction volume to trillions of transactions per second. So really prepare us for a high transaction sort of use case. But also, this Interplanetary Consensus will set a mode where we need to communicate and run blockchains across planets. imagine if any of our blockchain systems are only designed for running on Earth. But if we need to go to Mars, where the communication lag is huge, how do we make the consensus between two planets work in the solar system? So interplanetary consensus actually considers all these factors over huge distances over space, and they intend to make one single sort of network work. They can be used for many other projects. If you’re interested, please feel free to take a look at those projects.