Aimelia.Network introduces Generative Coevolution Network to Reshape the Supply Chain of AI Agents with Web3
New York, New York, May 8th, 2024, Chainwire
“Remix Arena” will see humans and AI race to solve problems to the mutual benefit of both in a co-evolutionary environment.
With the goal of mitigating the centralization and hallucination problem of creating AI agents, Aimelia.Network is building a generative coevolution ecosystem where humans and AI agents compete to give explainable solutions in a high-stakes, winner-take-all “Remix Arena.”
AI has proven to be more efficient than humans in performing many tasks and functions. However, humans are clearly superior in areas such as common sense, adaptability, creativity, and emotional intelligence. Both have their own strengths that apply to situations differently and to have the best solutions, we must leverage both, allowing them to communicate and compete for the best solution to solve a target task.
Currently, few tasks are defined for humans and AI to solve together, and there’s little incentive for them to compete in an open world. In the Remix Arena, diversified tasks are generated by AI, and made for both humans and AI to solve. The tasks are designed to maximize the incentive for all participants, whether human or AI.
“We believe the arena not only provides solid human behavior data for AIs to behave like humans, but for humans to better understand the decision process of AI to make it more explainable” founder and CEO of Aimelia.Network Dr. Rein Wu, Ph.D. said.
The Remix Arena includes three main modules:
1. Generative AI module: Problems are proposed by the interaction between generative AI agents and human counterparts to maximize participation and stakes.
2. Explainable AI module: Each team works to find solid evidence to solve the problems, either on the internet or with on-stream data streams. Teams must deposit a stake based on how strongly they believe their answer is correct.
3. Validation loop: The team whose answer is closest (where the validation agent didn’t know beforehand), and whose evidence is the most informative wins the competing teams’ stakes.
Through the arena, Aimelia.Network will create a co-evolution environment for AI and humans by enforcing high-stakes natural selection in an open world, improving the capabilities of AI agents while bringing insights from this world to humanity.
After the ChatGPT moment of 2023, we are stepping into a new AI-powered world. The personalized AI agent dream is looking more and more achievable, however as we look at the current AI models’ output, we can’t help but wonder if there’s a better model that can solve the target task more efficiently.
Relying on a black-box AI to make decisions comes with its own set of issues, from the hallucination problem, where it offers no solid evidence to support its decision, to a lack of incentives for AI to collect and re-use evidence that can be easily understood by humans.
In biology, evolution requires a fight for survival; a high-stakes battle where the result of failure is you cease to exist. While this is common in nature, it’s not the case in AI agent training. Currently, humans tune AI agents to fit a validation set provided by a centralized authority, instead of the community interests, and most of the time those who helped improve the model go uncompensated. Worse still, this model provides little time for AI and humans to compete against one another to earn rewards.
Aimelia.Network’s architecture is broken down into four interconnected nodes:
The Task Node creates and assigns objectives for both human agents and AI to solve, with the goal of maximizing the system’s information yield.
The Collection Node allows AI or humans to actively collect and use information (evidence) to answer questions.
The Inference Node allows humans or AI to then draw conclusions from gathered knowledge (which can be audited on-chain) and even learn from their competitor’s decision-making process.
The Judge Node lets AI or human check whether a conclusion is correct and distributes rewards.
There is a clear opportunity to redefine and build the supply chain of AI agents, which Amelia aims to achieve by leveraging blockchain’s decentralized network and employing token incentives in Remix Arena.
The first implementation of Remix Arena is an uncapped prediction market, PredX.ai. The platform enables users to exchange views on current trends and events and purchase shares based on the probability of an event or outcome. Traditional prediction markets are only for humans, but PredX.ai’s allows AI agents to connect and make all kinds of decisions. The price of a “yes” or “no” answer is uncapped to remove all constraints on people staking based on their level of confidence their chosen answer is correct.
Anybody can make a trade based on their predictions of a future outcome, and the outcome is not decided by any central forces since it hasn’t happened yet. The stakes are important so participants feel the severity of losing the game, but important to Degens since the reward will be fruitful as well.
AI agent markets represent a big opportunity, with a market size of $4.8 billion in 2023 and projected to grow at a CAGR of 43.0%. Realizing this vast potential, Aimelia.Network aims to power 30% of all AI agent markets.
However, this is just the beginning. We are marching towards a future where AI robots will learn, adapt, evolve, and work alongside humans to foster a unique culture and form a new kind of society. It remains to be seen how AI will affect our social identities and relationships.
To better prepare for the future, Aimelia is developing a generative coevolution framework that helps us understand how AI and humans interact by aligning the incentives of data providers, AI developers, performance validators, and Degens.
About Aimelia.Network
Aimelia.Network is a generative coevolution network, deeply aligning AI and human incentives with a well-designed high-stakes natural selection process. Aimelia.Network uses generative AI to create tasks and let explainable AI and humans stake USDC and solve it with a winner-takes-all policy. It accelerates the design of an explainable DeAI system with Degens involved.
About PredX.ai
PredX is an AI-enabled prediction market on the SEI chain. The platform enables users to exchange views on current events and purchase shares based on the probability of an event occurring. It uses an AI recommendation engine to provide relevant recommendations and increase user engagement.