Author: J.Hackworth, Variant data researcher

Translation: Golden Finance xiaozou

As more and more projects emerge in the Crypto x AI field, we are starting to see how encryption and artificial intelligence technologies work together on-chain.

In this article, we will explore how the potential of Crypto x AI is reflected in on-chain data.

Most on-chain transactions will be completed by AI agents. This trend is evident in the prediction market agents of Autoolas. Since the beginning of this year, the trading volume of agents has doubled. In the past month, the weekly trading volume has exceeded 38k, and approximately 63% of the transactions are Gnosis Safe transactions.

The Autonolas intelligent agent indicates that even with only about 453 intelligent agents running on-chain, their activities can generate more transactions than human activities from Gnosis Safe. Intelligent agents can achieve a more efficient future of task collaboration, and their output may exceed on-chain human activities.

Crypto is incentivizing the coordinated development of AI

The DeAI project is using encryption to guide new markets, attract talent and research funds. For example, Morpheus has about 6,000 addresses with a deposit amount of 93,000 stETH to increase liquidity and incentivize the development of artificial intelligence.

Token incentives are used to fund new applications/projects, create excellent models through competition, and improve model quality through RLHF. In addition to talent and skills, they also attract high-value assets such as computing resources and data.

The value of artificial intelligence depends on its training data. Crypto can enhance data collection through token incentives. For example, Hivemapper uses its network of over 63,000 users to collect map data while driving. Since early 2024, Hivemapper has burned tokens worth over $174,000 due to data demand.

Another example is rDAO, which incentivizes personal data. Reddit licensed user data to an AI company for over $200 million, but users may not have been compensated. In contrast, rDAO compensates users for providing data. So far, over 154,000 users have provided their personal data to rDAO for AI training.

AI reasoning is driving on-chain transactions, especially:

• Ora's AI oracle conducted 1700 transactions through opML.
• Io.Net's BC8 has had over 30,000 paid inference transactions on Aptos in the past month.

With AI making smart contracts smarter, transactions related to reasoning will rise.

More and more on-chain activities of AI indicate such a future:

• The intelligent agent will lead on-chain transactions
• New methods for stimulating the development of AI will emerge
• Compensate for new data
• on-chain reasoning rise