AI, Crypto, and the New Data Economy: A Conversation Between CZ and Vana

CZ on Vana Advisory Role

When asked about his decision to join Vana as an advisor, founder and former CEO of Binance, Chanpeng Zhao (CZ), explained that he has more availability to help strong founders right now, through his work at YZi Labs , noting that he can contribute valuable entrepreneurship experience while also learning from the team.

YZi Labs, formerly known as Binance Labs, is a venture arm where CZ works as an “intern”. Vana represents the fund’s first AI investment.

"I'm learning a lot about data annotation, labeling, training, data collection, etc. from you guys," said CZ, characterizing the relationship as mutually beneficial. He described asking "very novice questions" about how the system works, underscoring his philosophy of "not being afraid to be dumb" and asking questions when you don't know something.

"For me, it's a two-way experience," CZ continued, "I'm very honored to be here."

The AI Revolution and Blockchain

CZ describes artificial intelligence (AI) as a fundamental technology that's "not going to go away," comparing the current AI landscape to blockchain in 2009-2010. While blockchain has primarily focused on financial applications so far, he believes AI will have much wider applications beyond the current chatbots and language models.

"Crypto is the native currency for AI," CZ stated, highlighting that data is the fuel for AI models. He noted that AI has already "exhausted all the public data," creating a need for new data sources that aren't currently accessible on Web2 platforms.

Data Sovereignty: More Than Just Privacy

Anna spoke on how AI has transformed data ownership from a mere privacy preference to a significant economic question.

"In a world where AI models are extremely powerful and powering a huge part of the economy, representing a huge amount of revenue, then whoever owns that AI model is really aggregating a lot of economic power in the system," explained Kazlauskas.

She emphasized the difference between selling your data for a small amount versus trading your data for ownership in "the next ChatGPT."

CZ concurred, adding . "This is not just ownership of our data, but it's actually ownership of which data models you support", highlighting the ethical dimension of data control.

The Data Crisis in AI Development

Anna drew attention to a critical challenge: we're running out of training data for AI.

Leading AI models like Llama3 have been trained on approximately 15 trillion words or tokens, which is roughly equivalent to the high-quality content available on the public internet.

"We need somewhere to get more data," Anna explained. "When you have proper attribution and ownership, then you also are aligning incentives for people to share more data and actually scale up what's possible."

DataDAOs are a potential solution to this dilemma, providing decentralized training and datasets for AI.

DataDAOs: The User-Owned Solution

Anna describes DataDAOs as "a labor union for data where users can pool their data together to be able to create a valuable data set."

These data DAOs cover various types of information, from car telemetry and health data, to email data. Anna noted that the largest data DAOs have hundreds of thousands of users, with the overall ecosystem approaching one million users.

DataDAOs key innovation is non-custodial data, where users maintain full control of their data through private keys – similar to blockchain data. This stands in stark contrast to the Web2 model, where platforms effectively control user data.

Centralized data control leaves the door open for issues such as misinformation, information control, and manipulation.

Data as a Digital Asset Class

CZ emphasized that data is fundamentally a form of property that deserves protection.

Referencing Frédéric Bastiat’s 1850 book The Law, he stated that “Your data is your property… It needs to be protected by law or by technology and by both, really."

The former Binance CEO is a proponent ofVana's approach to creating DataDAOs where:

• Users maintain ownership of their data through private keys
• Data can be pooled to create valuable datasets for AI training
• Users can directly benefit economically from their data contributions
• Users can selectively choose which AI models to support with their data

In his closing remarks, CZ suggested that DataDAOs could even become a source of passive income for users. "Once you set it up, pretty much it's a passive income source... that only your own data have," he said.

Vana’s Roadmap and DataDAOs

When asked about Vana's future plans, Kazlauskas outlined an ambitious roadmap focused on scaling the ecosystem and its technological capabilities.

The immediate future will see continued development of existing DataDAOs, which already host hundreds of thousands of users collectively. The future of the ecosystem is set to include newDataDAOs for specialized data types such as Spotify listening history and Google location data through the Vana Academy and Vana Grow ecosystem programs.

Beyond simply growing the user base, Anna emphasized the importance of completing what she calls "the full lifecycle" of these data DAOs - moving from just data aggregation to actually training AI models and generating revenue from data sales. This next phase would represent a fundamental shift in how users participate in the AI economy.

As the initiative matures, Vana aims to reduce the amount of friction required to share data across the DataDAO ecosystem.

"We're working on programmatic data access," Anna explained, speaking on technical improvements that will allow developers to more easily access data across multiple DAOs without requiring manual approval for each use case.

Vana’s self-serve approach aims to preserve privacy while dramatically increasing usability for AI researchers and developers.

DeFi Meets DataFi

Anna emphasized that she's particularly excited about the “design space" that Vana is opening up for builders around financial primitives for data tokens.

Just as DeFi created an ecosystem of financial applications on blockchain, Anna shared that Vana is seeing early examples of "data-fi" emerge, including a decentralized exchange specifically for data tokens - DataDEX and even a platform for perpetual futures contracts based on data assets.

Future of Decentralized AI

The long-term vision is bold: Anna believes that within three years, Vana could help create ""a fully user-owned AI model, where you have 100 million users who all contribute data from five different data sources, and you actually beat the performance of some of the leading AI models."

This approach could serve as a mass onboarding event for Web3 adoption under the premise that "Your data is the capital".

"I see decentralized AI as potentially being the thing that actually brings the majority of the world on-chain," Anna concluded, suggesting that the path to self-sovereign finance might begin with self-sovereign data.

Learn more about Vana at www.vana.org