🧞Why crypto needs AI

And why AI needs crypto

Jun 17, 2024

GM and happy Monday frens🖖

AI is a major trend on the planet (ask Apple if you don't believe me) that will change everyone's lives, so the question of who and how will train, control models, data, personal assistants and autonomous agents is more important than ever.

So today's newsletter is about:

🕺 Must-see events
🤖 Crypto x AI use cases
📰 This week's news

👨🏻‍💻 dAGI House AI x Web3 Hackathon powered by cyber•Fund

Less than a month until the hackathon, where we'll build autonomous agents, interfaces for on-chain interactions, natural language or voice-enabled intents for wallet tasks, and use AI to enhance security and automate auditing. Join us right before EthCC to bring your ideas to life, win a share of the $100K prize pool, and launch your product.

Get ready for the main event in Brussels by joining our pre-conference online meet-ups.

More info here

⛓️L2con is soon...

Event for all people building on ETH & BTC with the best lineup ever. Here's a little sneak peek:

Mustafa AI-Bassam, Co-Founder & CEO @ Celestia
Jordi Baylina, Co-founder @ Polygon
Harry Kalodner, Co-founder & CTO @ Offchain Labs
Zac Williamson, CEO & Co-Founder @ Aztec
Erick de Moura, Co-Founder @ Cartesi
Anurag Arjun, Co-Founder @ Avail

And many more founders and leads...

🗓July 9, Brussels

🥩Option for restaking fans

Restaking & Infra Day is a one-day event for web3 developers, tech founders and non-technical web3 enthusiasts who want to dive deeper into the rapidly developing field of restaking technologies & solutions.

Listen to speakers from Eigen, Picasso, Etherfi, Trust Wallet, ZettaBlock and others!

🗓July 8, Brussels

Get me a ticket

🤿Deep-dive: Why AI needs crypto

Main insights from article written by our friend Stepan Gershuni, VC at cyber•Fund, firm investing in future of cybereconomy.

The original article envisions a cybereconomy powered by peer-to-peer networks, programmable markets, and AI, all designed to benefit humanity.

We collected 3 Web3 x AI use cases!

Zero-knowledge Machine Learning

Zero-knowledge technology, which allows for provably private computation, can also be applied to machine learning. This approach offers several advantages, such as performing inference on private data and making an ML model publicly available without revealing its code and weights.

Companies like EZKL, Giza Tech, and Modulus Labs are developing a zero-knowledge machine learning stack for practical applications in web3. Although zkML is not yet scalable enough to run large models like transformers in zk format, simpler ML algorithms like logistic regression, decision trees, and random forests can still greatly expand the design space for trustless applications.

Incentives Design

Blockchains are powerful systems for creating programmable incentives. They enable us to build networks that distribute rewards in programmable ways, expanding the possibilities for mechanism design. This technology can foster decentralized incentives for research and development in machine learning.

Take Bittensor, for example. It's a peer-to-peer network designed for collaboration on machine learning models and AI services. In this network, validators verify the work of miners and distribute rewards across subnets and individual miners. The goal is to produce the best possible solution to the machine learning task defined in the subnet code. Both miners and validators share equally in the rewards, which can be as high as $150,000 per day for each top-performing node (as of the time of writing).

Decentralized networks and organizations provide fair compensation to individuals working in machine learning research and engineering, free from the constraints of corporate guidelines and rigid hierarchies. The ability to create your own currency (token) can help kickstart the supply side of the network. However, in the long run, it will need to develop a sustainable token value accrual model.

Governance

The final, all-encompassing component of decentralized AI systems is governance. This raises crucial questions: Who decides what models and agents are permitted to do? Who owns and directs these systems? How is alignment ensured, and how is AI explainability managed?

Decentralized Autonomous Organizations (DAOs) are at the forefront of this new frontier, exploring the realm of programmable democracy. It's important to note that this isn't just a single project or framework. Instead, it's an entire design space encompassing everything from voting, execution, and treasury management to complete organizations that are augmented, enhanced, and managed by AI swarms.