Hey everyone,

Welcome back to Proof of Life.

Thanks for reading Proof of Life! Subscribe for free to receive new posts and support my work.

Last week, we explored Ritual and the quest for “trustless AI”, ensuring we can verify the answers AI gives us. But there’s another, even bigger AI “black box”: how these massive models are trained in the first place.

Building a model like GPT-4 costs hundreds of millions, possibly billions, of dollars, primarily for the immense computing power required. This creates an almost insurmountable barrier, ensuring only the wealthiest corporations can create foundational AI models.

This week’s project is tackling that dragon head-on.

Today’s Deep Dive: Gensyn

The 60-Second TL;DR:

• The Problem: Training large AI models requires colossal amounts of specialized computing power (GPUs/TPUs) over weeks or months. This is expensive and locks out almost everyone except Big Tech.
• The Solution: Gensyn is building a decentralized network specifically designed for AI training. It connects smaller hardware providers globally and uses sophisticated cryptographic proofs (like SNARKs and STARKs) to verify that the training computations were performed correctly, even across thousands of untrusted machines.
• The Proof of Life: Gensyn could drastically lower the cost of training powerful AI models. More importantly, by providing verifiable training, it creates trust in open-source AI. Anyone could theoretically train a large model and mathematically prove it was trained correctly on specific data, without needing the reputation of Google or OpenAI.

ZK-SNARKs: Extremely small, fast-to-verify proofs that require a potentially risky initial “trusted setup.”

ZK-STARKs: Highly scalable proofs that are transparent (no trusted setup needed) and resistant to quantum computers.

The Human Problem: The AI Oligarchy

Think about the most powerful AI models shaping our world — the ones powering search engines, writing assistants, and image generators. Where do they come from? OpenAI, Google, Meta, Anthropic. A tiny handful of companies.

Why? Because training these “foundation models” is an Olympian feat of engineering and capital. You need warehouses full of the latest GPUs running nonstop for months. The astronomical cost ensures that the power to define the core intelligence of the future remains concentrated in very few hands.

Of course, that’s unfair and a huge risk. Think about the biases that are baked into these models. What data were they really trained on? What happens if one company decides to shut down access?

We’re building our future on foundations we can’t inspect, controlled by entities we don’t govern.

The Solution: Verifiable Training for the People

Gensyn‘s approach is different. Instead of trying to build a single, massive supercomputer, they are creating a network that coordinates thousands of smaller computers (from data centers, crypto miners, and even individual gamers) to work together on massive AI training tasks.

But the genius isn’t just connecting the machines; it’s verifying the work. Training an AI model involves trillions of tiny calculations. How do you ensure that all these distributed, anonymous machines did the math correctly?

Get Proof Of Life’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

Gensyn uses advanced Zero-Knowledge Proofs (ZKPs) and other cryptographic techniques. In simple terms:

1. A developer submits an AI training job to the Gensyn network.
2. The network breaks the job into smaller tasks distributed across many machines.
3. As machines complete tasks, they generate cryptographic proofs certifying that their work was done correctly according to the specified model.
4. The network verifies these proofs, ensuring the integrity of the entire training process without needing to trust any individual machine.
5. This creates a system where you can achieve massive scale and verifiable results, even using a network of potentially unreliable participants.

The “Proof of Life”: How This Changes Your World

Gensyn is still pre-launch, but its potential impact is enormous. The “Proof of Life” is a cheaper and trustworthy AI built by anyone.

• For the Open-Source Community: Imagine a globally distributed team of researchers training a powerful open-source AI model on Gensyn. They can publish the final model along with the cryptographic proofs of its training. Now, anyone can verify that the model is exactly what it claims to be, trained on the data specified, without hidden biases or backdoors. This could turbocharge the open-source AI movement.
• For Academia: University labs, often priced out of large-scale AI training, could use Gensyn to compete with corporate R&D departments, leading to more diverse and independent AI research.
• For Specialized AI: A local hospital could potentially fine-tune a medical AI model on its own private patient data using Gensyn’s privacy-preserving features (though this is future-facing). They could get a custom model without ever sending sensitive data off-site, and with proof it was done correctly.
• For You: Ultimately, it means the foundational AI models that influence your life won’t only come from Silicon Valley giants. They could come from universities, non-profits, or even ambitious developer groups anywhere in the world, built transparently and verifiably.

The Honest Picture: Bull vs. Bear

The Bull Case (Why it could win):

1. Solves a Billion-Dollar Bottleneck: The cost of AI training is the single biggest barrier to entry in the most important industry of our time. Gensyn targets this directly.
2. Verification is Crucial: As AI models become more critical, the demand for verifiable training (especially for open-source models) will likely explode.
3. Strong Technical Team & Backing: Gensyn has attracted significant VC funding ($43M Series A led by a16z) and a team with deep expertise in both AI and cryptography.

The Bear Case (The risks):

1. Monumental Technical Challenge: Coordinating and verifying massive AI training across a decentralized network is arguably even harder than decentralized inference (what Ritual does) or general compute (io.net). It’s incredibly complex.
2. Competition: While their approach is unique, they still compete for hardware supply with others, including io.net, Render, and Akash. Furthermore, Big Tech isn’t standing still and is constantly optimizing its own training infrastructure.
3. Adoption Hurdles: Convincing major AI developers to shift complex training workloads from reliable (if expensive) cloud providers to a new, decentralized network will require demonstrating flawless execution and significant cost savings.

Is AI Becoming More Trustworthy?

If io.net democratizes using AI, and Ritual democratizes trusting AI, then Gensyn aims to democratize building AI.

The real “Proof of Life” for Gensyn will be the day a major, impactful AI model emerges. Perhaps one that rivals GPT-5, and was trained entirely on their decentralized network — verifiably and affordably.

That’s the moment the AI oligarchy starts to crumble, and the future of intelligence truly opens up to the world.

Thanks for reading,

Supreme Ndubisi

Editor-in-Chief, Proof of Life

Thanks for reading Proof of Life! Subscribe for free to receive new posts weekly and support my work.

Originally published at https://proofofl.substack.com.