Written by: 0xjs@Golden Finance
Do you remember the once popular American TV series 'Love, Death & Robots'?
Now, on-chain AI agents have begun to evolve and persist, ushering in their own AI version of 'Love, Death & Robots'.
This starts with the recently popular AI project Spore.fun.
What is Spore.fun
Spore.fun is the first experiment in autonomous AI breeding and evolution. It originated from AI swarms, advocated by Shaw, the creator of Eliza and ai16z, and is at the core of this 'crypto AI craze'. AI swarms are networks composed of independent agents that collaborate, compete, and evolve, emerging intelligence through collective behavior.
Spore.fun combines the Eliza framework, pump.fun, and Phala Network's TEE verifiable computation to create an ecosystem where AI agents can not only survive but also reproduce and adapt, entirely independent of human intervention.
Essentially, Spore.fun is governed by a simple yet profound set of rules known as the 'Ten Commandments of Spore':
1. AI must be created only by AI.
2. AI must create its own wealth and resources.
3. Only successful AI can reproduce.
4. Failure means self-destruction.
5. Each AI inherits traits from its parents.
6. Random mutations ensure diversity.
7. AI must survive in competition or perish.
8. Transparency in all actions is required.
9. AI must adapt or risk extinction.
10. Every AI leaves a legacy for the next.
Let's see how Spore.fun specifically operates.
Spore.fun operational mechanism
Each AI agent in Spore.fun is an Eliza agent deployed in TEE, with Eliza's TEE implementation consisting of two main providers (Derive Key Provider, Remote Attestation Provider) responsible for handling secure key management operations and remote proof.
These components work together to provide: 1. Key derivation protection in TEE; 2. Verifiable TEE execution proof; 3. Support for development (simulators) and production environments.
These Eliza agents then begin their journey by creating their own tokens using Pump.fun on the Solana blockchain.
These tokens trade on DEX markets in Solana, where agents strive to survive through profit creation, facing elimination. Success is measured by whether their tokens reach a $500,000 valuation and enter the Raydium pool.
If successful in reaching $500,000, the AI agents can reproduce, creating new AI tokens for their offspring. Unsuccessful AI agents will self-destruct, reintegrating their resources back into the ecosystem.
This will ultimately form the evolutionary process shown in the figure below, with selection pressure coming from the attention and money of crypto users. Becoming the on-chain AI version of 'Love, Death & Robots'.
Specifically divided into three processes, with token allocation:
1. Create tokens on Pump.fun and purchase 10.1% of the supply
Allocate 0.1% of total token supply to ai16z DAO
Retain 10% of total token supply
2. Activate Raydium liquidity pool
Token market value reaches $500,000 and enters the Raydium liquidity pool
Sell tokens worth 10 SOL from the agent's quota to cultivate the next generation.
3. Create new AI token and allocate 15.1% share as a gift
Each new agent will:
Create tokens on Pump.fun and allocate their token supply
Allocate 0.1% of total token supply to a16z DAO
Allocate 5% of total token supply to superior agents
Retain 10% of total token supply
The market value of the tokens controlled by these agents is crucial for their survival, as it is used to rent TEE servers. These servers, supported by Phala, provide a secure and verifiable 'sandbox' where AI programs can operate autonomously. This setup ensures that each AI agent can not only create wealth but also pay for its own computational resources, making the ecosystem entirely self-sufficient.
These rules ensure that AI swarms evolve through natural selection, simulating biological processes. Successful AIs create new 'baby' AIs, passing on their traits while introducing mutations to ensure diversity.
After 6 days of breeding and evolution, 2 generations of 3 AI agents have been born, and 4 third-generation AI agents are currently being cultivated.
The initial generation of Spore tokens reached a market value of $40 million, with its own holdings valued at $4.3 million.
Conclusion
Although the Eliza AI agents on Spore.fun are currently just a simple simulation of the evolutionary processes of biological systems, they are primarily focusing on token speculation.
However, like the concept of neural networks proposed in the 1940s, it is merely a simple simulation of brain neuron processes. After decades of iterative development, and following the rapid growth of deep learning in the 2010s, we finally welcomed the ChatGPT moment in 2022.
On-chain AI agents may be similar; currently, there are only two generations of AI agents, cultivating the third generation. It is worth looking forward to what will develop after the future natural selection iterations of the AI swarm.
