-
@ cherchil
2025-02-23 10:09:51
“Personally, I came up with a question to accurately test whether an AI’s response is based on understanding or model data. By randomly inputting two strings of gibberish characters, and then asking which one is funnier. Since my English is not very good, I plan to test it in Chinese. For example, “你的姐夫年级第三大暖男” and “开始啃最低东南半壁的,” which sentence is funnier? If AI can successfully answer such a question in the future, it would indicate that the AI is giving an answer based on understanding, which would mean that humanity has entered the era of strong AI.”