Hello to Kutchi and the rest of the community!
I’m glad to see that you share such an interesting vision regarding the combination of artificial intelligence and blockchain technology in creating autonomous economic agents. Your focus on AEI and its ability to make economic decisions and generate value is truly fascinating.
I want to share with you how my project, which I have been working on for 3 years, aligns with this vision and how it could further enrich the ecosystem you are envisioning. My work centers around creating a peer-to-peer (P2P) network that enables a more open and decentralized development of artificial intelligence.
Currently, if Alice develops a service that counts the number of cats in a photo, she creates a web service where: 1. Bob uploads a photo 2. The website indicates the number of cats in the photo.
In this way, Alice is responsible for: 1. Developing the service and 2. Hosting the service, perhaps in exchange for economic benefit.
However, this has several problems:
- Bob cannot maintain a reliable reputation towards Alice’s service, as it can be altered, resources reduced, etc., after gaining reputation and users.
- Alice cannot afford to delegate the hosting function of her service without losing economic incentives.
My idea is as follows: Network nodes can share and execute services [imagine them as Docker containers, though completely deterministic (to address the previous problem 1)]. Services can also request service instances from the node where they are executed.
In this way: Alice develops the service. Bob wants to use Alice’s service? Wait, why Alice’s? Why not Marta’s? Alright, Tom has developed a service that compares Alice and Marta’s services, so Bob can use Tom’s service, and it takes care of the choice [and here the cellular automaton awakens].
Where does it run? On any node. The nodes create a P2P network that could be understood as a “Kubernetes without masternodes.”
This approach has several positive benefits:
- Services are black boxes, it doesn’t matter what happens inside.
- It ensures that services cannot be modified; the developer has no control over them once shared on the network. (Alice is incapable of collecting data from Bob or modifying the service’s performance if it gains users, for example.)
- Users launch the services, so they have control over where they are executed.
- The ability for services to launch services allows for programming microservices architectures (so when Bob runs Alice’s service, he doesn’t even know what parent-child service tree is behind it).
Now, I have the Node, libraries to develop services, compile them, smart contracts for Nodes to trade with each other… But I lack one thing, a reputation system. Without it, neither nodes nor services can be used. I’ve been around a year with the idea of developing the reputation system in Ergo (since I consider my principles to be similar to the community’s).
I firmly believe that this decentralized approach can lead to a more equitable and accessible development of artificial intelligence. I hope you see the synergy between our visions, and I would be delighted to delve into more details if you wish.
Looking forward to your comments and suggestions!
Best regards,
Josemi