Gareth Jenkinson12 hours agoWill decentralized protocols become the future infrastructure of machine learning?AI and machine learning compute networks built on decentralized infrastructure could allow the onboarding of significant hardware resources worldwide.1100 Total views13 Total sharesListen to article 0:00InterviewOwn this piece of crypto historyCollect this article as NFTJoin us on social networksBooming interest in artificial intelligence (AI) and machine learning (ML) has led to a shortage of hardware resources and exorbitant cloud service costs, but decentralized infrastructure could challenge the dependence on centralized players.
Harry Grieve, co-founder of machine learning compute network Gensyn, spoke exclusively to Cointelegraph during the ETHGlobal event in London about the promise of peer-to-peer computing networks challenging Web2 services like Amazon Web Services.
Gensyn is an in-development decentralized network that will enable people to connect to various devices across the internet to train machine learning models. The company is backed by several Web3 venture capital firms and raised $50 million from Andreessen Horowitz in 2023.
Grieve says the network holds significant potential as the internet changes to a more dynamic representation of information that will empower “self-sovereignty and computational liberty online.”Grieve chats to Cointelegraph in London during the ETHGlobal hackathon. Source: Gareth Jenkinson
Gensyn has been in development since 2020, with Grieve and co-founder Ben Fielding researching machine learning computing for training and decentralized verifiable systems. The pair have been looking to solve a threefold problem with blockchain-based technology.
“How can you peer with another device and train a machine learning model on that device where A), the device is untrusted? B), your training model can’t fit on that single device. And C, you want the achievable scale of the entire system and unit economic outcomes as good as AWS,” Grieve said.
Gensyn’s lite paper describes the protocol as “a layer-1 trustless protocol for deep learning computation.” The network directly and immediately rewards participants for availing computing resources to the network and performing ML tasks.
Grieve said the challenge of building the network is verifying completed ML work, which sits at the intersection of complexity theory, game theory, cryptography and optimization:“We realized that if you want to build this in a decentralized way, you need a way to reach a decentralized consensus about who did what. That’s basically all a blockchain is.”Taking a leaf out of Satoshi’s Bitcoin playbook
Gensyn takes some inspiration from the ideals of the Bitcoin protocol, and Grieve says he’s a big believer in the old Bitcoin laptop mining days, where users could acquire BTC when it was still possible for smaller devices to be used for mining:“Satoshi gave people the right to their money again; they could generate it from a laptop. They could convert electricity into money. They could convert fiat into something harder.”
While the long-term plan is to make Gensyn a tool that allows a wide set of users and hardware to provide or access computing resources for ML training, the initial launch will target users with more graphics processing units (GPUs) because they represent the quickest way to get a lot of feedback.
“We are thoroughly planning on people building on top of Gensyn to make more user-friendly, supply-side, and contribution applications. Ultimately, an individual with a laptop will be able to download our client and run it in a way that connects you to the network,” Grieve said.
Related: io.net adds Apple chip support to its cloud-based GPU-sourcing network
Apple’s Silicon chips also promise to unlock massive global computing resources. Grieve says that research into Apple M2 and M3 chips shows that the hardware reaches parity with mid-tier, current-generation consumer Nvidia RTX GPUs.
This provides two potential major benefits for protocols like Gensyn, which could tap into a broad set of devices to contribute to its global supercluster. Grieve says:“I think people are more likely to have something that looks like a MacBook than a standalone GPU, so it’s a much more like democratizing force if you can contribute that.”
Apple Silicon chips are also highly versatile, acting as a “system on a chip” that could be emulated by other chip manufacturers.
Grieve believes the future could have more edge devices that are far more powerful than existing smartphones, adding that decentralizing across lots of devices and verifying device-agnostically is essential.
As Cointelegraph previously reported, Solana-based decentralized network io.net will onboard Apple Silicon chip hardware for its artificial intelligence and machine learning services.
Magazine: Is measuring blockchain transactions per second (TPS) stupid in 2024? Big Questions# Blockchain# Decentralization# Adoption# AI# Machine LearningAdd reactionAdd reactionRead moreFinal Fantasy 15 director says AI, VR, Web3 important for devs ‘whether you like it or not’Worldcoin makes Orb software open-source, implements ‘personal custody’Sleeper ICO-era project becomes 3rd most actively developed crypto app