根据UCC的一项研究, 人工智能写作显示出可预测的模式, 缺乏人类的风格变化. AI writing shows predictable patterns, lacking human stylistic variation, according to a UCC study.

大学科克学院的一项研究发现,人工智能生成的书写虽然流利、具有统一性、在文字选择和节奏方面具有可预测的模式,但与人类写作的文体变化和个性不同。 A University College Cork study finds AI-generated writing, though fluent, exhibits uniform, predictable patterns in word choice and rhythm, differing from human writing’s stylistic variation and individuality. 研究人员利用文学tyllogy, 分析了来自人类的故事和人工智能模型, 如GPT-3.5、GPT-4和Llama 70B, 揭示了人工智能文本与可检测到的算法来源“指印”紧密结合。 Using literary stylometry, researchers analyzed stories from humans and AI models like GPT-3.5, GPT-4, and Llama 70B, revealing that AI texts cluster closely together with a detectable "fingerprint" of algorithmic origin. 虽然人工智能可以模仿人类的连贯性,但它缺乏由个人经验塑造的创造性特质. While AI can mimic human-like coherence, it lacks the creative idiosyncrasies shaped by personal experience. 由于学生的写作自然不同,由于可靠性和伦理问题,研究人员告诫不要在学术环境中使用音量测量法来检测人工智能。 The researchers caution against using stylometry to detect AI in academic settings due to reliability and ethical concerns, as student writing naturally varies. 这些发现在生成性AI时代引发了关于真实性和作者身份的持续问题. The findings raise ongoing questions about authenticity and authorship in the age of generative AI.