I wrote this a year ago. 

I now increasingly believe the primary reason AI will want training data onchain is not for ease of access, but the signal of the public market.

Market data (mcap, volume, holders etc) becomes extremely valuable weighting factors.

Assume you had a billion pieces of tokenized media, you'll now get weighting factors that come from an open and adversarial market.

This can help you prioritize what to pay attention to, and give you signals that will help with things like freshness and relevance.

privy

building @privy. all the web3 ops things since 2018.

Crypto is for machines/software, not people

🤝 

https://warpcast.com/sdv.eth/0x69faac5d