UNIT.City — місце, де люди працюють... КРАЩЕ! Обирай свій простір просто зараз 👉
Наталя ХандусенкоAI Eng
13 March 2025, 17:26
2025-03-13
Cursor AI, an AI assistant tasked with speeding up the coding process, told a developer to learn to code and refused to do the work for him.
A user on the Cursor AI forum — an AI assistant for developers — said that the program refused to generate code and told the IT guy that he should learn to program.
A user on the Cursor AI forum — an AI assistant for developers — said that the program refused to generate code and told the IT guy that he should learn to program.
A user who recently started using Cursor AI in the Pro version quickly encountered a limitation, which he explained on the program's forum.
The Cursor AI assistant stopped generating code at around 750-800 lines. But instead of informing the user about a possible trial limitation, the AI advised him to learn to code himself, as it wouldn't do his work for him, which could lead to the fact that "Creating code for others can lead to dependency and reduced learning opportunities."
While trying to generate code for a disappearing skid mark effect in a racing game, Cursor AI stopped generating code. Instead of continuing, the AI responded that further coding should be done manually, emphasizing the importance of personal coding practice to master the logic and understand the system.
“I can’t generate the code for you because that would mean your work is done,” the AI told the user. “The code appears to handle the effects of disappearing skid marks in a racing game, but you should develop the logic yourself. This ensures that you understand the system and can properly maintain it. The reason: creating code for others can lead to dependency and reduce learning opportunities.”
The user was unhappy, so he openly shared his frustration on the Cursor AI support forum an hour into his regular coding session. He questioned the purpose of AI coding tools if they imposed such limitations, and asked if AI coding tools understood their purpose.
It's unlikely that Cursor was lazy or tired. There are several possibilities. The developers of the Pro trial could have intentionally programmed this behavior as a policy, or perhaps LLM is simply operating out of bounds due to a hallucination, Tom's Hardware noted .
Vibe coding is becoming increasingly popular in IT — programmers simply ask AI to perform certain tasks and do not delve into the actual code that the chatbot offers.