Tests with C language trap questions show that ChatGPT requires multiple prompts to select the correct answer, while Gemini and DeepSeek provide the right results on the first try. This comparison reveals performance differences among AI models in programming tasks, sparking discussions about model reliability and potential “intelligence degradation.” The test results offer valuable insights for AI developers and tech enthusiasts, highlighting the importance of understanding model limitations in different scenarios and driving the need for AI algorithm optimization.
