AI agents will be the rage.
The question is who will control them? In other words, what platform will they operate on? We have seen giants like Salesforce set a goal of millions of agents created for its platform. Other Big Tech companies are doing the same.
These agents are entering DeFi in a big way. The problem is that they are centralized, taking over the decentralized world. As more agents on centralized platforms operate, that swallows up a larger portion of the DeFi activity.
We are seeing these agents emerging from OpenAI or Anthropic. It is these entities that benefit from the data and transaction flow. Suddenly, that data is housed on major centralized clusters run by large technology companies.
AI agents are already here. Who will control them is what is uncertain.
Transparency was a promise of DeFi. When transactions are posted on-chain, using smart contracts tied to networks such as Ethereum, everything can be reviewed. There are no black boxes because it is all in the open.
We cannot say the same thing when smart contracts are built on Google or Amazon servers. Here we have non-verifiable smart contracts running, something that could pose a security threat.
Eventually, we could see the blockchain layer decentralized but the intelligence layer the exact opposite.
A parallel is the intelligence layer, via its coding, that Microsoft achieved. It is the operating system (intelligence) for desktop computers. These are manufactured by many different companies, existing throughout the world.
In this case, we could make the case the hardware is decentralized. No one company controls everything. The problem is that Microsoft is dominant at the intelligence layer. This has centralized to one company (two if we count IOS), at least on personal computers.
We see a similar situation with mobile. Apple and Google dominate the intelligence layer there. The phone come from a variety of manufacturers yet these two handle almost all of the volume for the O/S.
What we end up with is rentseekers. These entities extract enormous sums of money for their services. To illustrate we simply look at the conflicts regarding the app stores.
As always, decentralization is the answer. To achieve this, every layers of the stack has to focus upon distribution.
It all starts with infrastructure.
For the moment, the centralized entities are far ahead. This is common since these companies have tremendous resources to throw at the problem. We also see how they can resolve conflict simply due to the top-down structure. Once a decision is made, everyone follows.
Decentralized development is like herding cats. Ultimately, it takes time to reach consensus.
There is also a lot of "dot connecting". As more open source code appears, developers can start to incorporate more into what they do. This is a form of connecting the dots, creating larger programs based upon what is out there.
Centralized entities can plan their software stack. Each component is known by the organization (at least the upper level). Hence, the fitting it all together is incorporated from the start. While changes usually happen along the way, we see the focus targeted on the end result.
Decentralization is the opposite. We have infrastructure being built for AI yet it is still lacking as compared to the counterparts. This gap should close over the next few years as more awareness is brought forth.
People like Elon Musk consistently bring up the dangers of AI. Of course, people like him propose another centralized solution. Never do they suggest that we decentralize everything from the start. For those companies, it isnt a money maker.
Everyone needs to think about this. It is something that affects every human.
Those who simply default to centralized platforms or applications are putting everyone else at risk. It is the nature of where things are going.
Big Tech excelled at building a massive system which it captures most everything done in the digital world. However, individually, each of us can start to take basic actions to counter this.
Anything that is removed from Big Tech is a starting point. Each token of data that is captured elsewhere is a bonus. It starts the process.
As always, data is the food for these AI models. Proving all to Silicon Valley is not helpful.
