Vitalik Buterin called the possible use of AI for Ethereum
Ethereum co-founder Vitalik Buterin became interested in using AI for formal code verification and error detection.
"Probably the biggest technical risk for Ethereum right now is bugs in the code, and anything that could make a significant difference would be awesome," he wrote.
However, OpenZeppelin had previously received mixed results during experiments using GPT-4 from OpenAI to test the security of smart contracts in Solidity, the native Ethereum programming language.
At that time, the head of the company's machine learning division, Mariko Wakabayashi, noted in an article for Cointelegraph in July 2023 that the chatbot had identified vulnerabilities in only 20 of the 28 tasks assigned. According to her, in some cases the results were "frankly bad." Sometimes the neural network needed leading questions, and once it just invented a bug that did not exist.
Wakabayashi also recalled that Coinbase specialists received approximately the same results after experimenting with checking the security of tokens using ChatGPT.
"AI alone will not replace humans. However, auditors who learn how to use these tools will be much more effective than those who turn a blind eye to the new technology," she concluded.
Source: https://forklog.com/news/ai/vitalik-buterin-nazval-vozmozhnoe-primenenie-ii-dlya-ethereum
