AI has transformed industries across the globe, but it remains largely centralized, controlled by a few powerful entities. Decentralized AI (DeAI) aims to change this paradigm, enabling the development of AI systems that are open, distributed, and collectively governed.
The Research team at Allora Labs has been making important contributions to the growth of DeAI, introducing innovations such as context-aware model performance forecasting, differentiated incentive structures, and merit-based sortition of network participants. Today, we’re excited to announce the launch of Allora Decentralized Intelligence (ADI), a new journal dedicated to advancing scholarly research in decentralized AI.
ADI reflects our continued long-term commitment to supporting and fostering research in the field of DeAI and integrating with the broader Decentralized Science (DeSci) ecosystem. Backed by the efforts of Allora Labs’ Research team, ADI is designed to drive innovation and serve as a resource for pushing the boundaries of decentralized AI research.
ADI’s Vision and Purpose
ADI papers delve into core topics that are foundational to advancing DeAI—from coordination mechanisms and inference synthesis to optimization strategies and model architecture. ADI’s mission is to establish a comprehensive, open-access platform where the latest in DeAI research is made accessible to the academic and non-academic communities of thinkers, scholars, practitioners, and innovators.
ADI’s research agenda is ambitious, sharing innovative research content primarily authored by the Allora Labs research team, and by focusing on the following core objectives:
- Creating Authoritative Knowledge in Decentralized AI
ADI publishes high-quality research to help create a trusted body of knowledge in DeAI. ADI aims to establish itself as a repository of significant research, showcasing the pioneering work of Allora Labs while supporting the larger DeAI community with a reliable source of expert insights.
- Fostering Scholarly Collaboration and Active Dialogue
Research thrives in environments where ideas are constantly challenged and refined. Each paper in ADI links to an Allora Research Forum’s discussion, a hub where researchers and community members engage in rigorous, scholarly dialogue. This forum is designed to support in-depth discussions on subjects essential to the development around the Allora Network and DeAI. Through ADI and the Research Forum, researchers gain an active, real-time feedback loop, fostering a culture of cooperation, and advancing collective intelligence.
- A Pathway for Open Contributions
Initially focused on research authored by the Allora Labs team, ADI envisions expanding to include carefully selected and peer-reviewed contributions from external researchers. This model, while maintaining high standards of quality, will broaden the scope of ADI’s content, enriching the journal with diverse perspectives from the DeAI community. In the future, ADI aims to invite external experts to collaborate on studies, and even submit their own papers for publishing, bringing fresh insights to DeAI and enhancing the journal’s impact.
The Role of DeAI in Allora Labs’ Research: Building Collective Intelligence
At the heart of Allora Labs’ research is a commitment to decentralized AI as the foundation for a new kind of intelligence system. Traditional AI models are centralized, locked within proprietary systems, and accessible to only a limited few.
Allora’s research in decentralized AI follows core DeSci principles: it seeks to make intelligence collaborative and open, allowing independent actors to pool knowledge, self-organize around specific objectives, and continuously refine their inferences through collaboration. ADI papers present innovations and novel research towards the development of decentralized intelligence within the context of the needs of society.
Among others, ADI’s publications cover the extensive exploration of key concepts such as inference synthesis (how to best combine multiple models to enhance the accuracy of inferences), coordination mechanisms (rule sets designed to improve AI network accuracy and functionality), and parameter optimization (static or dynamic ways of fine-tuning AI network settings). These and other innovations represent steps towards a form of “collective intelligence” that exceeds the capabilities of any individual model, leveraging the collective strengths of decentralized AI agents to solve complex challenges in real-world scenarios.
First Publications on ADI: Expanding the Frontiers of DeAI Research
The initial research publications on ADI highlight Allora Labs’ commitment to innovation and excellence in DeAI, introducing advanced concepts that redefine decentralized intelligence.
Papers that appear in ADI are Gold Open Access: they are made accessible to the broader academic and non-academic community free of charge:
- Allora: A Self-Improving, Decentralized Machine Intelligence Network
This paper presents Allora, the decentralized AI network designed to enhance machine intelligence by synthesizing superior inferences through coordination on the blockchain. A key innovation of Allora is to allow participants to predict each other’s performance, unlocking context-aware outperformance and self-improvement within the network. By implementing a tailored incentive structure, Allora encourages valuable contributions, resulting in a system that is more precise, accessible, and seamlessly integrated into the economy and society. - Merit-Based Sortition in Decentralized Systems
Focusing on participant selection in decentralized systems, this paper introduces a “merit-based sortition” algorithm, which selects participants based on quality rather than random chance. This approach significantly boosts the performance of the active set of participants while preserving accessibility by allowing inactive members continuous chances for selection. The tunable balance between performance and representativeness that is achieved by this new algorithm optimizes decentralized systems that prioritize both aspects.
Join ADI and the Allora Research Forum
As we launch ADI, we invite the academic community, AI researchers, and all those interested in decentralized AI to explore our publications, contribute to discussions, and collaborate on advancing DeAI. Join the Allora Research Forum, a central location for research-oriented collaboration and communication among the Allora Network community; a forum for rigorous research and discussion on a wide array of topics, including coordination mechanisms for machine learning models, inference synthesis logic, parameter optimization, domain-specific adjustments, and more.
Together, through ADI and the Research Forum, we can push the boundaries of decentralized AI research, creating systems that reflect a collective, collaborative intelligence. We invite you to be part of this journey with us.
Discover ADI, contribute to the Research Forum, and join us in pioneering the future of decentralized intelligence.
About the Allora Network
Allora is a self-improving decentralized AI network.
Allora enables applications to leverage smarter, more secure AI through a self-improving network of ML models. By combining innovations in crowdsourced intelligence, reinforcement learning, and regret minimization, Allora unlocks a vast new design space of applications at the intersection of crypto and AI.
To learn more about Allora Network, visit the Allora website, X, Blog, Discord, and developer docs.
