Future Vision

Ungate aims to revolutionise the AI landscape by creating a decentralised, collaborative, and competitive environment for developing sophisticated AI systems. By leveraging swarm intelligence and a robust economic infrastructure, Ungate will facilitate the decentralised production and optimisation of AI models and agents. The vision for Ungate includes the development of SubNets, AutoNets, and SwarmDAOs, each serving distinct roles in enhancing AI capabilities and collaboration.

SubNets

SubNets are the foundational layer in the Ungate ecosystem. Initially, they will utilise industry-recognised AI benchmarking systems to evaluate and identify the best specialised models in various domains. Over time, SubNets will evolve to allow industry experts, researchers, and other stakeholders to create their own SubNets to evaluate AI models for specific areas. SubNets will:

• Utilise Established Benchmarks: Incorporate benchmarks like GLUE, SuperGLUE, ImageNet, COCO, MLPerf, SQuAD, SentEval, Librispeech, CASP, MLBench, and LMSys to ensure rigorous evaluation of AI models.

• Foster Competition: Create a competitive environment where models are continually tested and ranked based on their performance metrics.

• Enable Community Involvement: Provide tools and incentives for industry experts, researchers, and others to create new SubNets, fostering innovation and diversity in model evaluation.

• Enhance Model Selection: Provide users with access to the highest-performing models tailored to their specific needs, ensuring optimal performance and cost-efficiency.

AutoNets

AutoNets represent the next evolutionary step, focusing on the development and evaluation of specialised AI agents. Similar to SubNets, AutoNets will start with industry-standard benchmarks but will evolve to allow community-driven creation and evaluation. AutoNets will:

• Protocols and Tooling: Develop the necessary protocols and tools to create and manage AI agents.

• Competitive Dynamics: Use evaluation and competition dynamics to identify the best-performing agents, similar to how SubNets identify top models.

• Agent-Based Applications: Enable the deployment of specialised agents for various applications, enhancing the functionality and adaptability of the AI ecosystem.

• Community Involvement: Encourage the creation of AutoNets by experts and researchers to evaluate agents in different domains, promoting continuous innovation.

SwarmNetss

SwarmNetss will further extend Ungate's capabilities by facilitating collaboration among different actors to create Swarm Systems. These networks will:

• Collaboration and Governance: Implement DAO-like governance dynamics to manage the interactions and collaborations among different actors.

• Sophisticated AI Systems: Create complex, multi-agent AI systems capable of tackling advanced and multifaceted tasks.

• Innovation and Adaptability: Foster an environment where people and resources can collaboratively innovate and adapt to new challenges, pushing the boundaries of AI development.

Economic Infrastructure

To support these networks, Ungate will develop a comprehensive economic infrastructure that includes:

• Incentive Mechanisms: Reward model and agent providers based on performance metrics, encouraging continuous improvement and innovation.

• Transparent Transactions: Ensure all transactions are transparent and verifiable, fostering trust and accountability within the ecosystem.

• Economic Security: Extend the Actively Validated Service (AVS) to provide robust economic security around the evaluation and validation processes, ensuring the reliability of the system.