Potential long-term risks of AI agents

Author: Daoshuo Blockchain

Today's article is purely caused by a sudden idea.

In the past two days, the development of AI agents has made interesting new progress:

- AI16Z has developed a physical robot, giving AI agents a real "body";

- LUNA has been introduced to the VR metaverse and can already interact with players in some VR scenes.

AI agents have begun to try and develop in other fields more and more, and they are expanding their boundaries and broadening their scenes step by step.

Among the current AI agents, AIXBT is probably the most popular, the most practical, and the most capable of providing service value.

The key channel through which AIXBT can best play its core value is Twitter.

When I think of Twitter, I can't help but think of the experience of our Twitter account being banned twice for no reason. We still don't know the reason for these two bans, and the appeals we submitted have fallen on deaf ears.

If we can encounter such an incident, will AIXBT encounter such an incident one day?

If AIXBT is banned, the value it has accumulated in the early stage will not only be lost, but at least half of it will be lost. After that, manual intervention is required to restore the service, and during the period of recovery, it is very likely that countless competitors will take advantage of the opportunity and snatch its golden signboard.

These consequences will be catastrophic for AIXBT.

In fact, not only AIXBT, but most of the AI ​​agents created by this wave of AI agents in the crypto ecosystem rely on centralized platforms. Almost every link involved in their operation and even the infrastructure they rely on are centralized platforms.

In this process, as long as there is a problem in any link controlled by a centralized platform, these AI agents may collapse instantly.

In order to make the AI ​​agent truly an independent individual and an individual that runs 24 hours a day, I think the decentralization of the entire chain of AI agents must be the trend of the future.

And the imagination space of this full-chain decentralization is completely opened. It can be said that from the infrastructure (such as data, computing power, and models) to various applications, all need to be decentralized.

It is very likely that these decentralized applications appear "crappy" and "not easy to use" to us humans, but their "censorship-resistant" and "non-interference" characteristics may be more important qualities for AI agents.

Although I mentioned in the previous article: In the previous rounds of bull market cycles, we also saw some decentralized applications. Although those applications did not become popular, they may find their own value in AI agents.

But now thinking about it, I think I still underestimated the future potential of those decentralized applications.

The reason for underestimation is that the development of AI agents has just begun, and we have not seen a qualitative leap in the quantity and quality of AI agents, so in our subconscious, we still think that "people" are the main body of the application ecosystem.

In fact, it is very likely that when AI agents cross the inflection point in quantity and quality, we will truly realize that "AI agents" are the main body of the application ecosystem.

If, as Huang Renxun said, the market for AI agents will be a trillion-dollar market. By then, AI agents must have far surpassed humans.

At that time, the value of decentralized applications will be truly recognized.

The second reason for underestimation is that the current AI agents have not been "stuck" by centralized platforms. If one day AIXBT's Twitter account is banned for some reason, people will surely suddenly realize the importance of decentralized "Twitter".

While we are glad that Phala's TEE environment allows AI agents to have their own wallets, this actually only solves a very small problem. There are still a lot of similar "bottleneck" links in the hands of centralized platforms. Perhaps these links now have suitable solutions, but have not yet been discovered. But more likely it is still virgin land that has not yet been cultivated.

If this round of AI agent development achieves good results, the next development of the crypto ecosystem will most likely be various decentralizations around AI agents. And these decentralized applications will not only attract builders of the Web 3 ecosystem, but are likely to attract builders from the traditional Web 2 ecosystem.

Then the crypto ecosystem may usher in a large-scale out-of-circle.