ChatGPT from OpenAI has gone viral and reached 1 million users in just 5 days. The AI has shown mind-blowing capabilities, including generating smart contract code, and even fixing bugs for you. Is AI finally coming to put software developers out of a job?
AnChain.AI Web3 compliance officer and former regulator Michael Fasanello sits down with Senior Smart Contract Security Researcher Philip Werlau to dive into the power of ChatGPT, and discuss whether or not smart contract engineers should start dusting off their resumes.
We know that ChatGPT is capable of generating very convincing answers, but could it potentially do an even better job than a human? Watch the video to learn more and decide for yourself!
AnChain.AI Web3 compliance officer and former regulator Michael Fasanello sits down with Senior Smart Contract Security Researcher Philip Werlau to dive into the power of ChatGPT, and discuss whether or not smart contract engineers should start dusting off their resumes.
We know that ChatGPT is capable of generating very convincing answers, but could it potentially do an even better job than a human? Watch the video to learn more and decide for yourself!
Examples of ChatGPT being used to not only write code, but to detect vulnerabilities within that code have already generated significant traction on the web, but just how good is it at detecting vulnerabilities in smart contracts?
DeFi and smart contract hacks have already accounted for upwards of $3 Billion of lost funds in 2022 alone, with demand for experienced smart contract researchers skyrocketing to match. This has, however, created a significant squeeze on the already shallow pool of talent, with many firms waiting up to 6 months for a smart contract security audit.
Needless to say, if ChatGPT does have the ability to effectively detect vulnerabilities in smart contracts, it could accelerate time-to-market dramatically for would-be DeFi revolutionaries.
Unfortunately, though, it seems as if ChatGPT isn't quite ready for such an advanced use case. For our internal testing, it failed to detect vulnerabilities in all but 1 of our test contracts, despite all of them having known vulnerabilities.
It’s important to remember: ChatGPT is a static, non-adaptive language model that doesn’t check itself for correctness. It’s capable of generating very convincing answers, but if you’re asking it something you have no idea about, you could potentially get the wrong answer and not even know it. In this case, while ChatGPT can effectively explain the smart contract code in plain English, its ability to detect and fix vulnerabilities within smart contracts is limited.
Our Conclusion: For now, your jobs are safe! But it remains to be seen if the hardworking folks at OpenAI may yet have the last word on that...
DeFi and smart contract hacks have already accounted for upwards of $3 Billion of lost funds in 2022 alone, with demand for experienced smart contract researchers skyrocketing to match. This has, however, created a significant squeeze on the already shallow pool of talent, with many firms waiting up to 6 months for a smart contract security audit.
Needless to say, if ChatGPT does have the ability to effectively detect vulnerabilities in smart contracts, it could accelerate time-to-market dramatically for would-be DeFi revolutionaries.
Unfortunately, though, it seems as if ChatGPT isn't quite ready for such an advanced use case. For our internal testing, it failed to detect vulnerabilities in all but 1 of our test contracts, despite all of them having known vulnerabilities.
It’s important to remember: ChatGPT is a static, non-adaptive language model that doesn’t check itself for correctness. It’s capable of generating very convincing answers, but if you’re asking it something you have no idea about, you could potentially get the wrong answer and not even know it. In this case, while ChatGPT can effectively explain the smart contract code in plain English, its ability to detect and fix vulnerabilities within smart contracts is limited.
Our Conclusion: For now, your jobs are safe! But it remains to be seen if the hardworking folks at OpenAI may yet have the last word on that...
►Check us out on Twitter
►Questions? Check us out at https://anchain.ai or reach out to [email protected]
►Questions? Check us out at https://anchain.ai or reach out to [email protected]