研究表明，包含有问题的证据会使 ChatGPT 感到困惑，降低准确性。 Including evidence in question confuses ChatGPT, lowers accuracy: Study.

一项研究发现，在问题中包含证据会使 ChatGPT 感到困惑并降低其准确性，这表明没有证据的开放式问题可能会让 AI 模型得到更好的结果。 A study found that including evidence in a question can confuse ChatGPT and decrease its accuracy, suggesting that open-ended questions without evidence may yield better results from the AI model. 这一发现强调了问题构建对于从人工智能系统获得准确答案的重要性。 This discovery highlights the importance of question construction in obtaining accurate responses from AI systems. （来源：商业标准报、印度电讯报） (Source: Business Standard, Telegraph India)