At the Ethereum Community Conference (EthCC) held on July 19, Ken Timsit, Managing Director of Cronos Labs, presented an insightful analysis of the emerging use cases and limitations of blockchain-enabled AI. Drawing from research conducted on over 200 Cronos Accelerator applicants, He aimed to cut through the hype and identify high-impact applications in this domain. He also categorized crypto-AI innovation into three main areas: supercharging productivity, decentralized identity, and powering the AI economy.
Timsit emphasized that the most significant value in using AI for blockchain development lies in improving productivity and enhancing user experience. While the productivity levels in the crypto industry have yet to match those of traditional finance, AI can help bridge the gap. Marketing and customer support are particularly ripe for AI integration. For instance, Binance’s Project Kindred successfully generated free AI art for user profile pictures, serving as an imaginative marketing tool. Timsit also encouraged crypto projects to leverage AI for tasks like code development and testing. He highlighted that OpenZeppelin employs AI for code security, and while ChatGPT can auto-generate Solidity code, human oversight remains crucial. Developing developer tools that incorporate AI features is vital. Timsit explained that if blockchain developers wish to create effective tools, they must include products that facilitate their users’ utilization of AI capabilities.
He argued that although AI is reshaping the concept of truth and identity, it remains uncertain whether blockchain can effectively contribute to this area. While blockchain has the theoretical potential to track data lineage, credential issuers lack sufficient incentive to record information on-chain. Timsit explained, “When the blockchain community has attempted to address supply chain provenance, it has been disappointing because the parties generating credentials do not find economic motivation in putting those credentials on the chain.” Consequently, more research is necessary to develop decentralized identity solutions, especially considering AI’s capability to generate fake content. He concluded that the role of blockchain in this context remains unclear.
Timsit highlighted blockchain’s censorship resistance as a crucial factor for unhindered AI innovation, protecting developers from restrictive regulations influenced by tech giants. Also, crypto facilitates global monetization and commerce for AI models and services. However, running decentralized AI computing at scale still faces significant challenges. He noted, “Efficiently operating a decentralized GPU network for training models is an area where no one really possesses comprehensive knowledge.” Moreover, resolving hardware security concerns related to the private distribution of models is another hurdle that needs to be addressed.
Timsit expressed greater optimism about commercial applications and highlighted projects such as Fetch.ai, which employs crypto for paying for AI services between agents, and Cronos Labs Accelerator participant CorgiAI, which has established a marketplace for skilled engineers utilizing crypto payments. Overall, Timsit advocated for a practical approach that focuses on high-impact use cases, such as productivity gains. While powering the AI economy holds promise, further research and development are necessary to overcome existing limitations. Timsit suggested that blockchain’s most effective role may lie in enhancing existing systems rather than reinventing AI through excessive hype.