We are closing in on the era of digital labor. This is coming in the form of AI agents.
For those who regularly read these articles, this is no surprise. However, we are starting to see it gain a lot more traction in the mainstream. Events like we already had this year are only aiding in this endeavor.
Below is a video about a presentation put on by Marc Benioff, the CEO of Salesforce. He discusses how their platform already has thousands of agents working with thosuands of humans.
"I am not only managing human beings, I am managing agents".
This is digital labor. At present, we are dealing with cognitive but, as another CEO is showing, it will quickly move to the physical realm.
The image above is from a presentation given by Jensen Huang, CEO of NVIDIA.
We can see the different cycles:
The last one is self driving cars and robots. Nvidia announced a new platform for automakers to access, deriving synthetic data to train their systems. Before that, we can see what precedes it.
Of course, these will be happening simultaneously. The reason why this is likely structured in such a way is the fact that bits are faster than atoms. We are looking at a couple years for robots to emerge in any number. Data collection on this end is still slow.
Companies are already starting to integrate agents into workflow processes. As we move through 2025, this will continue.
At the same time, we are going to see Big Tech adding this to their platforms. This will obviously help the customers of companies such as Salesforce and Amazon, where business use their cloud services.
Here is where we come to the decentralized versus centralized question. What is really going to be open and available to the masses?
Taking this from the data perspective, the data feedback loop only expands at more agents are created. Each interact generates synthethic data, something that is exponentially higher than humans can create.
To top it off, things really accelerate when AI agents interact with each other autonomously. Consider all the engagement a "digital workforce" could produce.
Which harkens the main question: where is the data?
Under the described scenario, it is in the hands of Big Tech. Anyone using platforms that Salesforce owns is placing it on their servers. This is the situation with cloud. This will allow Salesforce to train models, which are provided to their customers.
The same holds true for Nvidia. And Google. Microsoft. Apple. Amazon.
Are these the companies we want controlling the next generation of "workers"?
They will not be the only game in town however. The decentralized community needs to keep pushing ahead, replicating what Big Tech is doing.
We have promoted the idea of democratized data as the crossroad humanity coming to. Everything discussed regarding AI, from the Big Tech perspective, is around data. The main focus is acquiring, utilizing, and generate more of it.
This needs to be the focus of Web 3.0.
Fortunately, there are some steps being taken. Each person that is involved should be seeking out alternatives.
Some are in the process of building out service that can access what is out there. Raw data is the starting point but structuring it is also important.
RSS3 has introduced a comprehensive Web3 dataset on Hugging Face, serving as a valuable resource for AI applications. This decentralized and verifiable dataset transforms machine learning by granting developers open access to structured data derived from platforms such as Farcaster and Lens, fundamentally reshaping the construction of AI and decentralized applications.
There are basically two kinds of unlimited data: synthetic and real world using cameras. This is the next wave that companies must access.
AI agents are synthetic generators. Imagine a digital workforce of 25 billion agents. There will come a time when this is reached. How much data is going to be generated at that point?
In a couple decades, the economy will not resemble what we have today. We are going to see a massive transformation in economic output.
Data centers are becoming factories. Instead of manufacturing a phyisical product, they are producing virtual ones. The result will be cognitive output, workers, and automated products and services. This aligns with generative, agentic, and physical AI.
The entire population of the planet is roughly 9 billion people. Consider the impact of 25 billion agents, many of them tied to physical machines. What does the economy output look like.
Of course, this will bring the age old debate about the owners of the means of production to the center of all discussions. The present path is leading up to one where Big Tech owns it all.
Digital labor is coming. Are the humans ready?
