To enhance the cost-effectiveness of AI inference, it might be necessary for projects to develop their own blockchains. Here’s the rationale:

Open-source AI models, such as Llama3, are becoming widely accessible and commoditized. These models can be used directly for inference or serve as the basis for AI agents

Pro decentralization but also believes that road goes through sudo decentralization.

If a company operates these models internally, it can guarantee the reliability of both the execution and the results.

Conversely, using a third-party service requires trusting that the execution is accurate. This issue of trust is especially critical in blockchain applications where there are financial implications.

Currently, platforms like Chainlink facilitate the output of AI models on the blockchain, but the main obstacle is confirming the accuracy of these executions.

This is where zero-knowledge (zk) technology becomes vital. Since zk technology is costly on current blockchains, launching zk-friendly chains or rollups that can efficiently verify zk proofs could reduce costs.

This scenario opens significant opportunities for integration between the web3 and web2 ecosystems, capitalizing on the synergies between AI and blockchain technologies.

Let me know if you disagree and why?