A quick look at the current status and future of AI Agent

Author: jolestar

I was playing around with AI Agent last week, and I attended ai16z’s event in Beijing the day before yesterday. I wanted to see what AI Agent can actually do now, and think about what it can do in the future.

The current state of AI Agent reminds me of that meme about a person hiding in a vending machine. The AI ​​Agent that everyone has imagined has begun to have autonomous consciousness, but in fact, there is actually a developer hidden in the AI ​​Agent. (Everyone is making up the picture here. I tried to let AI generate this picture and found that AI cannot understand "hide")

How the AI ​​Agent framework basically works

The AI ​​Agent framework currently plays the role of a glue, gluing the client (Twitter, Discord, Telegram, etc.) and various plug-ins (chains, etc.), and then the framework provides a basic library (memory storage, session isolation, context generation ), etc., and will be connected to various AI platform interfaces later.

How the AI ​​Agent framework is integrated with applications and business

scenarios

Since AI became popular last year, various platforms and tools have emerged. The most critical thing is to solve a problem, how to combine AI with applications. Some AI platforms try to provide plug-ins, some create workflow models, and some traditional applications embed AI within the application. But the key here is: 1. Where is the interactive entrance of the application? 2. How to integrate AI with existing business logic.

The application interaction portal provided by each AI platform to users is a dialog box similar to a chat window. Obviously everyone believes that the interaction method with AI applications should be a "personification" method. The smart thing about AI Agent in this regard is that it directly connects to all open IM and social systems, which is obviously easier to accept than building a new one.

How to combine AI with existing business logic. The solution provided by AI Agent allows developers to integrate AI decision-making into business scenarios. Programming languages ​​need to be deterministic. The condition of if can only be true or false and cannot handle fuzzy business logic. Through AI, complex logic can be converted into precise conditions, which can then be seamlessly integrated into business scenarios.

For example, the function of replying to messages in a group needs to be triggered by traditional IM Bot through some clear message instructions, but through AI, a method shouldReplyMessage can be implemented, and given context, it returns true or false.

The main functions of AI in business logic scenarios are:

1. "Intention" discovery: Through the description in the prompt word, let AI discover the "intention" in the user's text message based on the context, and map the intention to a specific code.

2. Assist decision-making: Use AI to convert fuzzy and complex conditions into definite true/false or enumeration types, and then integrate them into business logic.

Seeing this, many people may be disappointed with AI Agent. Many people think that AI Agent is just teaching AI and it can do everything. In fact, due to the context limitations of large models, there is no way (at least currently) to create a universal AI that can do anything. But the good news is that programmers don’t have to worry about losing their jobs. There still needs to be a large number of programmers hidden behind AI, and someone still needs to stack if elses. But the key difference is that the business boundaries that the program can handle are expanding.

Two AI Agents

At the event, I asked shaw a question. The market has two expectations for AI Agent. 1. AI Agent plays a role by itself, has its own ID and brand, and provides services to users. 2. The user has a personal AI Agent, which is equivalent to a personal assistant and can assist the user in handling some business. Which of these two AI Agents will be more popular? He felt that both directions would be good, and it was possible to combine them.

Now on the market, what everyone is mainly exploring is the first direction. This direction is similar to the service AI Agentization. In the future, there may be no App interface. Apps will become AI Agents and anthropomorphic. The second direction is the agentization of the application client. In the future, the application client will be a plug-in of the assistant Agent. The application local data becomes part of the Agent memory library. At the same time, this plug-in is also responsible for communicating with the cloud service Agent. And this is a new application architecture model that will change the entire infrastructure.

AI Agent infrastructure requirements

1. The infrastructure must be permissionless, otherwise the AI ​​Agent will be restricted by various anti-attack strategies, and the service should use economical cost (Gas) to prevent attacks. At this point, platforms with a relatively low degree of openness will face a greater impact, and the open platform craze in the early days of Web2 will be rekindled.

2. AI Agent needs to be able to operate funds to pay to solve the above problems.

In other words, future services, whether based on blockchain or not, need to support Crypto's private key mode authentication and Crypto-based payments.

The combination of AI Agent and chain

In addition to the two points mentioned above, how to combine AI Agent with the chain is a direction that everyone is exploring. At the event, chat with Mikkke about the focEliza he is working on. Of the two AI Agents mentioned earlier, at least the first one requires a running or verification environment provided by the chain. Because once an AI Agent provides services to the outside world, there will be trust issues, and its role is actually the same as a smart contract.

There was a controversy at the time about the name "smart contract". It was just a piece of code. What was "smart" about it? AI can make smart contracts worthy of their name. The difficult problem is how to call the AI ​​interface in a smart contract environment. If it is still a long way to go to allow large models to run in a verifiable environment, using a solution like Oracle is a more feasible path.

There will be a lot of demands surrounding AI Agent. How to obtain the public knowledge of AI Agent? How does AI Agent determine facts? How does AI Agent identify the same user on different platforms? How is "memory" stored in smart contracts? If I have multiple devices, each with an AI Agent installed, how do they share memory?

You will find that the "data uploading", relationship uploading, DID, P2P network, etc. that were previously done in Web3 have new meanings and scenarios.

Conclusion

Reusing the conclusion I shared about AI and blockchain once in 21 years, an Internet that is more friendly to AI is also an Internet that is more friendly to humans. At that time it was just a thought, but now the future is here.