FLock partners with Io.net to launch “Proof of AI” to secure and verify decentralized networks

One of the biggest challenges we face online is verifying the authenticity and quality of the content we consume, whether it’s news, videos, images, or social media posts. With so much information out there, it’s challenging to separate fact from fiction and credible sources from unreliable ones.

This is where decentralized content verification (DCV) into play. DCV uses blockchain technology and artificial intelligence to create a transparent and secure system for verifying online content. Its goal is to help users, creators, and platforms build and maintain trust in the digital space. Two key players driving the DCV initiative are io.net, the world’s largest Decentralized Physical Infrastructure Network (DePIN), and Flock.io, a decentralized AI training platform based on federated learning.

Today, io.net and Flock.io announced a partnership to develop the world’s first Proof of AI (PoAI)—an AI-native Proof of Work and consensus mechanism designed to validate the integrity of nodes operating within a decentralized compute network, potentially shaping the future of DePIN projects.

With PoAI, decentralized physical infrastructure networks (DePINs) can verify the integrity of their nodes by requiring them to complete compute-intensive AI training tasks. PoAI is an AI-focused Proof of Work that directs verification efforts toward meaningful AI tasks. This allows nodes to earn block rewards from DePIN as well as AI training networks, specifically through both io.net and Flock.io.

The mechanism, which authenticates DePIN nodes in a decentralized and AI-native manner, involves an engine that continuously presents challenges, gathers responses, and provides necessary stats (e.g., latency, score deviation, data correctness) to io.net nodes to form assessments.

“AI engineers and end users need to trust the quality of the compute resources available to them, and Proof of AI is crucial in making that happen,” said Jiahao Sun, Founder and CEO of Flock. “Compute drives the entire AI development process, so we’re starting there, and we’re excited to collaborate with io.net, a leader in this space.”

Decentralized AI ecosystem providers can use PoAI to confirm the honesty of participants in their networks and ensure high-quality AI training inputs. Flock is responsible for providing continuous challenges, such as generating synthetic data through inference on a large language model (LLM), that dictate expected behavior from GPU devices based on factors like TFLOPS and VRAM.

The PoAI service will create a self-sustaining cycle that links Flock’s decentralized training platform with decentralized AI DePIN platforms. Over time, Flock plans to extend PoAI’s verification capabilities beyond compute tasks to include AI training and federated learning, with discussions underway about the possibility of a dedicated AI chain.

“The introduction of Proof of AI will bring significant improvements in AI model training and inference over decentralized compute networks,” predicted io.net CEO and Co-founder Tory Green. “I’m confident GPU node operators and the broader AI/ML developer community will embrace Proof of AI enthusiastically—we certainly have.”

While synthetic data has proven extremely useful for model training, synthesizing and cleaning vast amounts of data, such as the 15 trillion tokens used in LLama3 training, is a complex task. As a result, Flock Data Generation will utilize idle GPU resources to perform batch inference on LLMs as requested by Flock Task Creator and Training Node.

Decentralized AI GPU networks are essential for the long-term success of decentralized AI, but dishonest actors still attempt to exploit the system. A common tactic is tricking the network into believing they possess more computing power than they actually do.

Without strong deterrence measures, node operators could act dishonestly to secure network rewards without making meaningful contributions. Solving the node integrity verification issue has been challenging, as bad actors can misrepresent their resources and collect rewards without performing the necessary work.

With Proof of AI, Flock and io.net believe they have achieved a significant breakthrough, laying the groundwork for future advancements in decentralized AI.

Flock.io is a decentralized AI training platform based on federated learning. The platform democratizes model creation, development, and alignment by incentivizing the community to contribute training data and model feedback. Protocols and developers can use Flock to train models tailored to their specific needs, such as transaction agents, AI companions, function call models, health analytics, and more.

Launched in November 2023, the genesis of Io.net stemmed from a pressing demand for vast computing power, particularly in executing intricate algorithms for trading stocks and cryptocurrencies. Recognizing this niche, the founder of Io.net embarked on a mission to establish a decentralized network of GPUs capable of meeting the escalating requirements of AI and machine learning (ML) enterprises.

Fast forward to today, Io.net boasts an impressive network comprising over 25,000 nodes, delivering access to hundreds of thousands of GPUs for AI/ML ventures. This innovative approach has not only slashed costs but also significantly reduced lead times compared to conventional cloud services.

FLock and Io.net Partner on “Proof of AI” for Decentralized Compute Verification