probably?
even if the LLM is trained on flawless C code (which it isn't) it still has no way of reasoning about a complex system, it's just "what token is statistically most likely to come next"
even if the LLM is trained on flawless C code (which it isn't) it still has no way of reasoning about a complex system, it's just "what token is statistically most likely to come next"