New Stack Overflow rules: it is forbidden to use ChatGPT to answer questions. Violators can seal for up to 30 days.

CTOnews.com, December 12, OpenAI recently launched a chat AI, ChatGPT, which can generate code, fix bug, consult online, and even answer a variety of questions, causing netizens to compete for experience. Stack Overflow, a well-known programmer question and answer website, previously announced that it would temporarily ban the posting of ChatGPT-generated text on Stack Overflow.

Now, the Stack Overflow has officially upgraded the interim rule to a new community rule and has proposed disciplinary measures that can ban violators for up to 30 days.

Stack Overflow officials say Stack Overflow is a community based on trust. The community believes that the answers submitted by users reflect the exact content they actually know, and that they and their peers have the knowledge and skills to verify these answers. The system relies on users to use official tools to verify the contributions of other users, including the responsible use of yes and no votes.

Currently, the contribution generated by GPT usually does not meet these standards and therefore cannot contribute to a trusted environment. This trust is broken if the user copies and pastes the information into the answer without verifying that the answer provided by GPT is correct, ensuring that the source used in the answer (services not provided by GPT) is correctly referenced, and that the answer provided by GPT answers the question clearly and succinctly.

The objective nature of the content on Stack Overflow means that if any part of the answer is wrong, the answer is objectively wrong. In order for Stack Overflow to maintain a strong standard, such answers must be edited or replaced as a reliable source of correct and verified information. However, because GPT is enough to convince site users that the answer is valuable, the signals commonly used by the community to determine the legitimacy of peer contributions often fail to detect serious questions about answers generated by GPT. As a result, objectively wrong information will enter the website. In its current state, GPT may undermine readers' trust in the site to provide answers written by subject matter experts.

CTOnews.com learned that Stack Overflow pointed out that moderators have the right (at their own discretion) to ban users who copy and paste GPT content on the site for up to 30 days, with or without prior notice or warning.