March 16, 2012 - Foreign media outlet Ars Technica reported this Friday that a man who uses a Cursor AI developer "janswist" of the racing game project was writing code using Cursor.Programming AssistantSuddenly "quit" and offered unsolicited career advice instead.
According to a bug report on the official Cursor forums, after generating about 750 to 800 lines of code (which users refer to as "locs"), the AI Assistant stops working and sends a rejection message: "I can't generate the code for you because that would be generating the code for you! Getting the job done ...... You shouldDesign your own logic so you can really understand the systemand ensure smooth follow-up maintenance."
AI even provided a paternalistic justification for the developer, stating that "generating code for others mightLeads to dependency and reduced learning opportunities”.
janswist filed a bug report directly on the company's product forums, spouting, "Cursor I can't believe you're making me learn how to program., instead of generating code for me." And a screenshot was attached.
janswist guessed that heProbably hits some sort of upper limit on Cursor code generation.The AI is not a good choice, but it's about 750 to 800 rows. However, other users have reported that when they use the Cursor, the AI Still generates more codeOne commenter suggested that janswist should try Cursor's "Agent" mode because it is more suitable for large programming projects. One commenter suggested that janswist should try Cursor's "Agent" mode, which is better suited for large programming projects.
Note: Launched in 2024, Cursor is an AI code editor based on an external large-language model similar to those supporting generative AI chatbots, such as OpenAI's GPT-4o and Claude 3.7 Sonnet.It offers code completion, interpretation, refactoring, and full function generation based on natural language descriptions, and is rapidly gaining popularity among many software developers. The company also offers a Pro version, which is claimed to have enhanced features and greater code generation limits.
Not coincidentally, Dario Amodei, CEO of Anthropic, recently made the point that future AI models could beEquipped with an "exit button", to opt out of tasks they find unpleasant.