“Skullface sends his regards: Empirical evidence of Large Language Model's influence on human spoken communication”
Empirical evidence of Large Language Model's influence on human spoken communication
ChatGPT偏爱的词汇如delve,在人们口语中突然变多
作者发现 ChatGPT 偏好生成的词(如 delve、showcase、boast、intricacies、meticulous)在人类自发口语中骤增。他们对 824,634 期播客、共 737,083 小时对话做了合成控制分析,把这一变化因果地与 ChatGPT 的发布联系起来;一项预注册实验(N=496)也证实,短暂的聊天机器人互动就会让参与者采用它的词汇,并在干扰任务后仍然保持。作者认为这说明在人类数据上训练的机器正把自己的特征反馈到人类语言中,并带来语言同质化和少数 AI 提供者潜在文化影响力的担忧。
HN 评论者反复把它当作“人们开始像 ChatGPT 一样说话、LLM 词汇渗入口语”的证据,并常用来调侃别人的措辞受了 LLM 影响。
HN 书评
推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及“people are already beginning to talk like and/or think like chatGPT:”
“Also, monkey see, monkey speak: Empirical evidence of Large Language Model's influence on human spoken communication”
“I don't need to "prove it", because all I have to do is link this:”
“Also, even if you somehow didn't AI generate this, you sure as shit got infected by the LLM mind-virus and now you write/talk like it.”
“The extremely charitable take is that they got infected by the LLM mind-virus: Empirical evidence of Large Language Model's influence on human spoken communication”
看全部 7 条 →
“The linked study used audio of academic talks and podcasts to determine this.”
- Empirical evidence of LLM's influence on human spoken communication ↗ransom_rs · 2025-07-13 · 6 分
- Empirical evidence of LLM's influence on human spoken communication ↗pseudolus · 2025-07-16 · 3 分