With $35 million in funding, Allora, led by veteran VCs like PolyChain and Framework, has recently been particularly eye-catching. I've seen many people refer to it as a "prediction market." That's incorrect. Let me explain my understanding of this project:

1) To be precise, Allora is a decentralized AI inference service platform. Users with any AI judgment needs, including price prediction, strategy optimization, risk assessment, etc., can pay for services provided by AI Agents. Therefore, the prediction market is just one application scenario of Allora, not the entirety.

2) Given that AI models inherently have varying inference output capabilities, how can they become mature upstream suppliers for mass output? The answer lies in Allora building a collaborative and competitive aggregation platform powered by AI models.

Its mechanism is straightforward. For example, if a user wants to predict whether ETH will rise or fall and how to set the LP price range, the traditional approach is to look at candlestick charts and listen to KOL analyses or purchase various customized AI model APIs for predictions, only to find a bunch of differing answers. Can there be an aggregation inference service platform to handle this comparison and selection process?

The key is this: after users submit their requests to Allora, the network architecture with 280,000 nodes will compete to provide answers. Some will say it will rise, some will say it will fall, and some will say it will consolidate. Allora will vote on these models and record historical performance, giving higher weight to AI models with a high prediction success rate and sending token rewards, while penalizing those that guess randomly.

This creates a positive feedback loop: accurate models earn more, gain higher weight, and take on more tasks; those that keep guessing randomly are eliminated.

3) Therefore, I tend to view Allora as the infrastructure layer for AI inference services, capable of on-demand invocation of AI model combinations. There are mainly two major application scenarios:

DeFAI: When AI Agents execute on-chain transactions, they need to determine whether a transaction is subject to MEV, provide the optimal price range in real-time when adjusting Uniswap LP, assess whether AAVE has liquidation risks, and dynamically adjust leverage rates for Yield pools, etc.;

Prediction Market: Using AI models to dynamically adjust and update probabilities, compared to mechanisms that purely rely on trading volume for pricing, AI's aggregated inference can provide users with a more intelligent starting point for predictions, avoiding purely following the crowd.

However, fundamentally, Allora is still just an infrastructure service. In the early stages, with fewer models and less data, and insufficient accuracy, it will also experience a long energy accumulation period.

But if in the future, both DeFAI and prediction markets can become mainstream, the value of its infrastructure services will become prominent.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。