Thinking about the Concept of Web4: Agents augmented Liaison
The term “Web3” once became a buzzword and was widely popular. However, with the significant drop in the prices of cryptocurrencies closely related to Web3, and the emergence of other buzzwords, its presence in the news has diminished. Nonetheless, the concept and services of Web3 are still in the process of spreading and evolving technically.
Meanwhile, the term “Web4” is also starting to emerge in discussions, particularly in the context of the European Commission advocating for the concept of Web4.
There isn’t a concrete manifestation of Web4 yet; it’s primarily a term referring to concepts and possibilities. Based on my personal understanding, I would like to reflect on the concept and history of the Web and contemplate my interpretation of the Web4 concept.
Concept of the Web
The terms “Internet” and “Web” may seem similar but have slightly different meanings.
The term “Internet” refers to the communication network used for activities like publishing and viewing blogs, using social networking services (SNS), and streaming services. As a general term, it encompasses the services and cultural and communication activities on the Internet.
“Web” is short for World Wide Web. You might have noticed “www” at the beginning of website URLs, which stands for World Wide Web. Due to the lengthiness of saying “www,” it has become common to abbreviate it to “Web.”
Technically, while the Internet signifies a communication network, the term Web is more ambiguous and broad. It generally refers to the common technologies and services provided primarily through web browsers and web servers, assuming the Internet as the underlying foundation.
In general terms, the Web also represents the services and cultural and communication activities facilitated through web browsers and web servers.
Thus, the Web typically refers to technologies and services accessed via web browsers, not dedicated apps. Technologies confined to client-side devices like PCs and smartphones or skewed towards data centers and cloud services are not central to the Web. Likewise, common technologies biased towards applications are not the essence of the Web.
Of course, the scope of the Web extends beyond this, but at its core are the web browsers and web servers and the associated technologies and concepts.
Evolution of the Web’s Social Function
With its evolution, the Web has provided social common functions.
Web1 was about individuals disseminating information through web pages, making information broadcasting, previously dominated by mass media, accessible to individuals.
Web2 brought about interpersonal bidirectional remote communication through SNS, enabling low-cost digital communication compared to the relatively expensive and analog methods of postal and telephone communication.
Web3 is about decentralized management of information and infrastructure and system realization. In essence, it allows for various services on the Internet to be realized without relying on major IT companies. The value of this function may be less apparent in societies where individuals’ freedom and privacy are protected and large companies tend to act ethically.
However, if we assume scenarios where companies providing various Internet services misuse our information or spread biased information, the decentralization feature of Web3 becomes significantly meaningful.
Personally, I don't believe that preventing such misuse or abuse through Web3 technology is necessarily the optimal solution to these problems, as they also exist outside the Internet. Many of these issues are better resolved socially. Nonetheless, having more options is preferable, and I think it's about finding the best mix of social systems and technology.
From Web1 to Web3, the social function has been to provide media for our communication and information dissemination activities, enabling cheap, easy, extensive, and safe communication.
The Liaison Function Aimed by Web4
When considering the social function of Web4, I believe the keyword is the liaison function.
Liaison refers to the role of actively bridging gaps or barriers between individuals or groups to promote the construction of smooth and good relationships.
Web4 should aim to provide this liaison function to society, as it is feasible with modern technology. I believe contemporary society is eagerly awaiting such a manifestation of Web4.
There are three aspects to this liaison role of the Web:
1. The first aspect is the elimination of language barriers through automatic translation technology and natural language processing AI. This vision includes not just translating words but also transcending barriers in specialized fields, culture, customs, and ways of thinking, playing a liaison role through the Web.
2. The second is a more proactive liaison. This involves connecting people engaged in related research, adjusting to reduce the likelihood of arguments, and envisioning the Web as an active lubricant in human relationships.
3. The third aspect is the liaison between humans and technology. Web3 was complex and not easily approachable. Also, various technologies and social systems are complex, making it challenging for individuals to grasp all the details.
Conversational AI and technologies emphasizing intuitiveness and ease of understanding can create a foundation where many people can easily use and try out advanced features. This could be the image of Web4.
Technologies for Realizing Web4
The core technology for these liaison roles is natural language processing AI. The emergence of conversational AI, made possible by recent large-scale language models, allows us to concretely envision the form of Web4 as a liaison.
Here, AI technology carries the core function of liaison, but the evolution of Web technology is also necessary to widely offer this to society. This technological evolution will be the essence of Web4.
For AI to be integrated into society as a liaison, it is necessary to operate various AI agents that take on different liaison roles in our devices and cloud servers in data centers.
These AI agents should be able to communicate with us, naturally support our activities, and collaborate with each other. The ability to develop, provide, deploy, and utilize various AI agents on this common functionality is what I envision as the technological aspect of Web4.
The Image of Agents
Thus, I believe the keyword for the technical aspect of Web4 is "agent."
Currently, programs known as bots are active on the Internet. However, bots are not viewed very positively, as they often put a significant load on networks and servers and are used for illicit activities or spamming.
If we can curb these negative images and problematic uses, and make it easy for anyone to use agents as they wish, agents, including bots, could become incredibly convenient.
The advent of language AI, which allows instructions in natural language, has paved the way for everyone to master agents. Current chat-type AIs only respond to messages and do not perform agent-like operations.
However, technically, creating a program that operates in the background upon receiving instructions and continues to process those instructions is not very difficult. Linking this with AI could easily create a basic AI agent.
Technical Challenges of Agents
The integration of numerous AI agents into the current internet environment is not without its challenges, including cost and infrastructure issues.
First, the autonomous operation of numerous agents could significantly burden the network and servers, leading to enormous costs and the risk of overloading the infrastructure capacity, potentially causing outages.
When humans use the internet, even if they access many servers simultaneously, there is a limit, and the frequency of access is not very high. However, since agents are automated programs, a single computer could, in theory, execute processes that impose thousands of times more communication and load on the network and servers than a human.
If every internet user were to do this, the internet would quickly become congested, rendering most functions unusable.
Therefore, communication protocols and management systems that can handle the communication and loads of numerous agents, addressing load and security issues, are necessary.
Moreover, technologies to enhance the reliability of agents, to allow them to be operated with confidence, are required.
Ideas for Addressing Technical Challenges
For communication protocols and management systems suitable for the age of agents, I am personally working on designing a protocol called ASOS (ASynchronous Object System).
ASOS aims to be a Web protocol that minimizes the amount of messages necessary for information gathering and communication among a vast number of autonomous agents while ensuring security and managing communication limits and billing.
Regarding the reliability of agents, I am envisioning the concept of meta-agents.
This idea involves creating ultra-reliable AI agents that are validated through an open, decentralized system like Web3, gathering legitimacy verification results from many people. These highly reliable AI agents would be meta-agents.
Meta-agents would verify the source code and operation of other regular agents, ensuring that the agents function as intended and are free from malice or vulnerabilities. This would allow us to use publicly available agents freely without the burden of technical knowledge or verification. Additionally, agent developers would be able to freely develop and release new agents.
With these mechanisms to optimize the load caused by agents and technologies to use agents with confidence, the era of Web4, where free coexistence with agents is possible, will arrive, and Web4 will be able to provide various liaison functions to society.
Our Information Risk
We store our information on devices in hand, on servers in the cloud, and on blogs and social networking services (SNS). We also provide personal information to many internet service and public infrastructure providers.
These practices expose us to risks concerning our privacy and safe living. Sometimes, they even lead to risks of financial loss or becoming victims of crime.
Even now, information leaks from companies occur frequently. Data stored on cloud services or in various providers' databases is constantly exposed to these risks. Additionally, with the advent of quantum computing in the future, even encrypted data may face increased leakage risks.
In the Case of Agents
Reflecting on this situation, it seems preferable to avoid unnecessarily exposing information and only provide it when necessary, asking companies to immediately dispose of it. If information is only provided when needed, there's no need for companies to store it indefinitely.
It's impractical for humans to respond every time a company needs our information due to the inconvenience for both parties. However, with agents, human effort is unnecessary. If our agents and the company's agents are verified at the source code level by meta-agents and pass the necessary assessments, such an arrangement becomes feasible.
We would simply need to register a simple ID with various service providers and give them the contact information for the corresponding agent. There's no need to register credit card numbers, addresses, phone numbers, age, gender, or even names. For example, when I shop using my ID at an online store, the company's agent could request payment from my agent. The purchased items are then handed over to the delivery service, along with the contact information of my agent. The online shop company does not need my credit card number or address.
My agent, authenticated by a meta-agent, will securely process the payment from my bank account or Web3 wallet. I cannot deny or alter the operation of my payment agent program. The company can await payment processing with peace of mind after conducting credit checks with my agent.
By querying the contact information attached to the package, the delivery service can deliver it to my address.
If the online shop's or delivery service's agents are not authenticated by a meta-agent, my agent will not pass on the information. Agents of duly authenticated companies will use my information only for the duration necessary for their operations and then immediately dispose of it.
Thus, companies can trust that I will pay without knowing any of my personal information. Regardless of the payment, agents will assure various aspects of my situation on my behalf to third parties. Personal information like addresses will only be known to those physically needing it for their operations, and only for the duration of those operations.
This is an example of a society with agents and the vision aimed for by Web4.
Liaison through Agents
In addition to guaranteeing trust in business transactions and protecting personal information, agents also play the liaison role I mentioned as part of the Web4 concept.
Agents can passively observe our online communications and internet searches. Currently, it's possible for companies providing various online services to assist our activities with their support AI. However, it's challenging under current conditions for us to receive support from our chosen agents for each online service.
The proliferation of standard communication protocols that allow for monitoring and intervention by agents will make such activities easier. This is also part of the vision for Web4.
My chosen agent, for example, could monitor my communications on social networks or messengers, translating languages if necessary, or converting specialized or niche terms into more understandable language. It can also tactfully smooth out any language that might hurt the other party or make me uncomfortable.
Even in heated discussions, the agent can organize the points of debate and devise ways to facilitate calm and constructive discussions before distrust escalates or exchanges become defamatory.
For tasks involving complex technology, agents can provide comprehensive support, creating or advising on what we desire.
In this way, agents with various abilities and characteristics will adeptly support our everyday activities within the natural detection range of online activities and smartphones. This is the ideal image of the liaison function in Web4.
In Conclusion
In this article, I explored my ideas about the concept of Web4.
It's unclear whether a reliable method to ensure the trustworthiness of the key AI in meta-agents will be found. Additionally, it's uncertain whether conversational AIs can develop the ability to understand the nuances of human relationships and provide support without hindering activities.
On the other hand, most of the other aspects seem to involve technologies that are largely realizable. Not just limited to my personal research on ASOS, but suitable protocols and platforms for agent communication will likely be realized eventually.
Of course, even if all these are in place, it does not mean that the risks associated with agents disappear. There is no way to completely prevent the misuse or abuse of technology or accidents due to errors. Therefore, even with the advent of Web4 and meta-agents, we must continue to address security issues diligently.
