- The latest version of the artificial intelligence chatbot, ChatGPT v4, is capable of identifying vulnerabilities in Smart Contracts, or smart contracts.
- Former Coinbase head Conor Grogan embedded an Ethereum smart contract into GPT-4 and the chatbot identified “security vulnerabilities.”
The latest version of the artificial intelligence chatbot, ChatGPT v4, is much more powerful than the previous version, so much so that It is capable of identifying vulnerabilities in Smart Contracts, or smart contracts.
GPT-4 is the latest milestone of Open AI that, “while less capable than humans in many real-world scenarios, he exhibits human-level performance in various academic and professional benchmarks”.
In fact, it is now capable of converting image, audio, and video input to text, as well as handling “much more nuanced instructions”.
ChatGPT v4 passes the SAT exam
This March 14 were published the scores obtained by the new version of the chatbot under test. He was found to have passed a bar exam with a score in the top 10% of those tested. For comparison, the previous version was “in the bottom 10%”.
What OpenAI does is put the chatbot to perform a series of exams and, for example, in the LSAT test, which must be taken by anyone who wants to be admitted to law school in the United States, ChatGPT obtained a score of 163 in the 88th percentile.
This means that, if it could, ChatGPT could be admitted to one of the top 20 law schools in the United States. While, in the previous version, the chatbot scored 149 in the 40th percentile. So the progress is noticeable.
Clearly, ChatGPT is a tool that, although not yet close to full human capability, is extremely useful for many different tasks, and apparently, GPT-4 is capable of auditing smart contracts.
GPT-4 could reduce vulnerabilities in smart contracts
Smart contracts, although they are a fundamental part of the crypto industry, include a set of vulnerabilities that endanger the ecosystem.
It is necessary to remember that a smart contract is a series of lines of code with many conditionals. Obviously, human error may be present or, conversely, vulnerabilities may be intentionally included.
precisely for this reason It is so important that people make sure that the smart contract of a crypto project has been audited by a reliable third-party company. But even so, the risk does not disappear because there is always room for human error.
However, this could be about to end. Former Coinbase head Conor Grogan said on Twitter that he embedded an Ethereum smart contract into GPT-4 and the chatbot identified “security vulnerabilities.” In fact, he described how he could exploit the code.
Conor added that, interestingly, precisely that contract was hacked in 2018 through the vulnerabilities found by ChatGPT.
“I believe AI will ultimately help make smart contracts more secure and easier to build, two of the biggest impediments to mass adoption.” concluded the former director of Coinbase.
As if there wasn’t enough proof of the new potential of ChatGPT, Rowan Cheung, founder of The Rundown, shared a video on Twitter showing how GPT-4 transcribed a hand-drawn website into code.
Advances in technology always raise concerns, especially in terms of security. However, history has shown that progress is inevitable.
You might be interested in: