Decentralized artificial intelligence infrastructure firm Ritual and blockchain-based federated AI model training project FLock.io, announced a partnership today to improve the development of decentralized AI models.

As more AI companies join the industry, decentralized AI has become a trending topic as companies such as OpenAI and Google LLC dominate headlines. Decentralized networks allow users to train and deploy models that users can maintain control over without having to hand over control of their data or models.

Ritual is a Web3 network platform that uses blockchain technology to allow traditional and blockchain-based businesses to deploy AI models on its infrastructure to execute AI tasks, known as inference. Since it’s blockchain-based, it’s decentralized with an open AI infrastructure that supports blockchain smart contracts. Those are pieces of self-executing software that can natively connect to AI models.

FLock.io complements this by providing a decentralized machine-learning platform that allows AI developers to train and refine their AI models. This is the first step before deploying models for apps. It uses federated learning, a type of machine learning that uses multiple machines to train a model, and blockchain technology to provide secure environments for model training while maintaining data privacy.

Unlike centralized commercial and open-source model training, FLock’s decentralized training uses multiple user-contributed nodes, which split training up into parts, which then are synchronized into one global model. Users who contribute models, data or computing resources for nodes with “FLock Points” for participating in AI model customization. These points incentivize participation with potential rewards such as crypto tokens and other benefits.

The two companies said this partnership will give developers who use Ritual’s inference ecosystem to run their models access to FLock.io’s capability to fine-tune them to fit their needs for specialized AI solutions. For example, the capability to customize models to answer specific questions about finance, healthcare, retail or other industry-specific data.

FLock.io said it is exploring the use of Ritual’s blockchain capability to store key data used in model training such as model parameters and scoring on-chain, as well as incentive reward distribution information used for contributions. That ensures that contributors and developers can view the full lifecycle of AI models and incentives.

Image: Production Perig/Shutterstock