“Subligence”——提议用于大型语言模型(LLM)“智能”的新词。

3作者: akiarie大约 1 个月前原帖
叫我雪花也好,但我提议那些不认为人工智能是真正智能的人,寻找一个术语来指代它的“思维”,而不是“智能”,因为这有助于我们脆弱的思维理清事理。 **subligence /sʌbˈlɪdʒəns/ n.** 一种较低或初步的智能形式;在没有完全理解的情况下做出反应、选择或适应的能力。 动物、机器、系统或无生命物体中似乎存在的思维。 在理性之下的辨别能力;低级智能。 例句:“自去年十二月以来,编码模型的表现明显变得更加低级,以至于未加辨别的眼睛很难将它们的输出与人类智能区分开来。”
查看原文
Call me a snowflake, but I propose that those of us who don&#x27;t believe that AI is actually intelligence look for a term to refer to its &quot;thinking&quot; that isn&#x27;t &quot;intelligence&quot;, because this helps our weak minds keep things straight.<p>subligence &#x2F;sʌbˈlɪdʒəns&#x2F; n.<p>A lesser or rudimentary form of intelligence; the capacity to respond, select, or adapt without full understanding. The semblance of mind in animals, machines, systems, or inanimate things. Discernment beneath reason; inferior intelligence.<p>Example: &quot;Since December, the coding models have gotten noticeably more subligent, so much so that the undiscerning eye has trouble distinguishing their output from human intelligence.&quot;