The Optimist - ORA Protocol - Onchain AI Oracles - Revelo Intel

The Optimist – ORA Protocol – Onchain AI Oracles

In The Optimist’s Twitter Spaces which took place on June 17, 2024, Charles hosted Kartin to discuss AI oracles, the integration of AI and crypto, and more! Read our notes below to learn more.

Background

Charles (Host) – Co-Leader at The Optimist

Kartin (Guest) – Co-founder of ORA Protocol

ORA Protocol – a verifiable Oracle protocol that brings AI and complex computing on-chain.

Exploring AI Oracles: Kartin from ORA Protocol 

  • Kartin introduces himself as the co-founder of ORA Protocol. He shares that he got into crypto through Bitcoin and started a mining program in high school. After a brief hiatus to focus on astronomy research and working at Google, he returned to crypto in 2022. Inspired by DeFi summer, Kartin started ORA Protocol, noticing the need for better Oracle solutions, especially with the rise of AI.
  • Kartin explains that traditional oracles, like Chainlink, are not efficient for AI computations due to their high cost and complexity. He and his team developed a new approach called Optimistic Machine Learning (opML) to create a more cost-effective and scalable solution for AI oracles. This involves using optimistic rollup concepts to verify AI computations on the blockchain.
  • Kartin explains that an AI oracle focuses on performing AI computations off-chain and verifying them on-chain. Traditional oracles rely on majority voting to determine the correctness of data, which is too costly for AI computations. AI oracles use opML to allow anyone to perform the computation and submit the result, with a fraud-proof mechanism to ensure accuracy. This makes AI oracles more suitable for complex tasks such as generating images with AI models or serving as financial judges in DeFi applications.
  • Kartin gives an example of how an AI oracle works, mentioning a scenario where someone requests an AI-generated image with a bounty. The result is verified through a decentralized process, where anyone can challenge and prove if the computation was not performed correctly. This system can be used in various applications, including financial decisions, NFT generation, and on-chain gaming.
  • Kartin highlights that AI oracles enable permissionless, decentralized computation delegation, which is crucial for integrating AI with blockchain technology.

Revolutionizing Oracles with AI: ORA Protocol and Future Use Cases 

  • Charles asks if multiple LLMs are being used in the ORA Protocol and how LLMs and opML combine to create the protocol.
  • Kartin explains that ORA Protocol uses various AI models and combines them with opML for decentralized and verifiable AI inference. He mentions the importance of a verifiable mechanism to ensure the integrity of AI inference, which traditional methods lack. opML allows anyone to execute AI inference jobs and submit them for verification on the blockchain.
  • Kartin explains that the ORA Protocol can support many AI models, allowing decentralized AI inference. He mentions that the market is still unclear about the best practices, so they are exploring different approaches. 
  • Kartin envisions the protocol being used for automation, cost efficiency, and supporting various models. One feature in development is continuous execution without the need for repeated prompts, which would save on gas fees.
  • Charles asks about the Initial Model Offering (IMO) of ORA protocol. Kartin explains that IMOs make AI models ownable by everyone, similar to ICOs but focused on individual models rather than companies. 
  • Each model can have its own IMO, making it a distinct and ownable asset. AI oracles can generate significant profits, especially in sectors like lending and market making. For instance, an AI model could manage a lending protocol, analyzing data to determine collateral rates and other parameters.
  • Kartin also mentions the potential impact on NFTs and on-chain gaming. As AI-generated content becomes more prevalent, the focus will shift from human-centric to AI-centric content creation. AI-generated NFTs and on-chain games could reshape these sectors, offering new opportunities for profit. He highlights that AI will transform various aspects of the crypto space, and there will be interesting speculation about which models will succeed financially.

Shaping the Future of AI and Crypto: Initial Model Offerings

  • Kartin explains that AI and crypto are both in paradigm-shifting stages, where AI will change crypto, and crypto can also change AI. He stresses that tokenization is the first and easiest step in this transformation. He points out that while tokenizing physical assets faces enforcement challenges, virtual products are more suited for tokenization.
  • Kartin explains how the success of OpenAI can be attributed to its ability to work without the pressure of immediate returns, allowing researchers to focus on long-term goals. He hopes the Initial Model Offering (IMO) can create a similar environment for AI development, allowing researchers to crowdfund their models and share ownership with investors.
  • Kartin describes two types of IMOs: one for existing models and one for models yet to be developed. 
    • The first type involves tokenizing existing models, making them ownable, and allowing investors to share in the profits. 
    • The second type involves crowdfunding for new models, where researchers can raise funds based on their reputation and plans for innovative AI architectures. Investors in these IMOs own part of the model and benefit from its success.
  • Kartin highlights the example of the open language model (OLM), which is fully open-source and permissionless. He explains that various research teams have bought $OLM tokens and are improving the model, as they now have a financial incentive to do so. He mentions a new product built by the OLM research team that uses on-chain Oracle to provide AI services, making it unstoppable and fully decentralized.

Tokenizing AI: How ORA Protocol is Transforming Investment

  • Charles summarizes that using the open ML on ORA Protocol allows users to invest easily via tokenization by buying or earning tokens. These tokens represent ownership in the AI models, and users can generate fees from them. If users have insights into AI models or oracles, they can invest in models they find interesting or useful in DeFi.
  • Kartin explains that the point system is straightforward, involving simple tasks like following on Twitter and using the AI model on-chain. Users who delegate AI requests to OAO (on-chain AI Oracle) will be eligible for future airdrops through the point system.
  • Kartin highlights that ORA Protocol is a permissionless Oracle protocol, meaning users can claim points even if they use products built on ORA without knowing it. This seamless integration allows users to benefit from the platform’s rewards without direct interaction.
  • Kartin says that AI will make the crypto industry more capital-efficient and liquid. He expresses a preference for AI-managed compounds, which provide liquidity for any token asset on the blockchain. The better the asset, the higher the rate it will be able to collect.

AI DeFi and Future Innovations: ORA Protocol 

  • Charles mentions the challenges of training AI models for risk management to avoid issues like bad debt or illiquid collateral in DeFi protocols. He suggests that while AI has the potential to manage these risks, it requires careful development.
  • Kartin agrees and speculates that while this summer is focused on meme coins, the next major trend could be AI DeFi, following the previous trends of ICOs and DeFi summers. He explains that AI can significantly enhance liquidity and capital efficiency in the crypto industry.
  • Charles believes that AI represents a major UX improvement for DeFi, enabling users to interact with agents that handle complex tasks, simplifying the user experience. He sees this as a significant advancement that many did not foresee.
  • Kartin encourages users to support open source AI by purchasing $OLM tokens and participating in the ORA ecosystem. He highlights the importance of involving researchers and users in building a transparent AI future on the blockchain.
  • Kartin also recommends users explore ecosystem projects like 7007 AI, an inference asset protocol.

Check out these important links

Show Information

  • Medium: Twitter (Audio)
  • Show: The Optimist Twitter Space 
  • Show Title: Ora Protocol – Onchain AI Oracles
  • Show Date: June 17, 2024